HEMISPHERIC LATERALIZATION OF JAPANESE KANJIAND KANA:EVIDENCE FOR RIGHT HEMISPHERE INVOLVEMENT
IN SEMANTIC PROCESSING OF KANJI
B. A. ,Doshisha UniversityM. Ed. ,M. A. ,Northeastern University
by
Submitted in partial fulfillment of the requirements for the degree of Doctor ofPhilosophy in Psychology in the Graduate School of Arts and Sciences of Northeastern
University, January, 1990.
Chisato Aoki
.
ABSTRACT OF DISSERTATION
The purpose of this research was to investigate whether the right hemisphere of
the human brain is involved in linguistic (semantic) processing. Among right-handed
persons, linguistic processing has been shown to be lateralized primarily in the left
hemisphere (LH), while visual-spatial processing has been thought to be lateralized
primarily in the right hemisphere (RH) (e. g. Gazzaniga, 1967; Kimura, 1961, 1966).
Japanese orthography is a good tool with which to study this general issue because it
has both ideographic characters, Kanji, and syllabic characters, Kana. Access to the
Kanjilexicon may be achieved via visual-orthographic codes, without phonological
mediation (Sait0, 1981), and so may involve the RH. However, access to the Kana
lexicon requires prelexical phonological processing (Gory0, 1987; Sait0, 1981),
presumably involving the LH. Thus the hypothesis of this thesis was that the RH
would be better than the LH in processing the meaning of Kanji words, but the
opposite pattern would hold for Kana words.
Previous research has generally supported LH processing for Kana and RH for
Kanji, although the pattern of results does depend on the task. For Kanji, earlier
results are generally in line with the expectation of RH dominance for visually-
sensitive tasks (Hatta, 1977; Sasanuma at a1. , 1977). Only when high-level semantic
tasks are used, such as superordinate Gategorization and judgment of semantic
congruence, does LH dominance occur with Kanji(Hatta, 1981; Hayashi & Hatta, 1982;
Sasanuma at a1. , 1981). However, lateralization of the comprehension of word meaning,
a lower-levelsemantic task, has not been previously tested. It is interesting to test RH
involvement in word comprehension in normals, partly because clinical observations on
Japanese partial split-brain and alexic patients suggest that the RH may be able to
understand familiar and concrete Kanji words. The relative dominance of the RH and
LH was tested in a series of tachistoscopic half-visual field experiments.
A picture-word matching task was used to test the hypothesis. This task
minimizes overt phonological processing, and does not require higher cognitive
processes such as superordinate categorization. Two sets of experiments were
conducted using this task. The first set (Expts. I-4) employed both Kanji and Kana.
The pictures and words were matched for meaning (the "semantic" task). The reaction
time (RT) for making the match was recorded. A cross-over interaction for the RTS
between character type and visual field (i. e. , hemisphere) predicted by the hypothesis
was obtained in a clear-cut fashion. This result extended previous work (Jones & Aoki,
1988).
In the second set of Experiments'(6-8), only Kanji characters were used. These
were tested not only in the "semantic" task but also in a "phonological" task in which
the sound of the word was matched to the sound of the picture's name. Again, RTS
showed an interaction between task and visual field, providing additional evidence to
support the hypothesis that the RH is involved in comprehending Kanji words. This
interaction, though clear in the first session, tended to wash out over subsequent
sessions..
For both Kanji and Kana, the predicted interaction was strong only when both
the picture and word were presented to the same visual field. When the picture was
centered and only the word was lateralized, the interaction was reduced or disappeared.
These aspects of the data may be consistent with hemispheric activation theory
(Kmsbourne, 1973).
Taken together, the results of this research indicate that in the picture-word
matching task, using highly familiar and highly concrete words and lateralized
presentations of both the picture and the character, Kana words are processed better in
the LH, while Kanji words are processed better in the RH. This pattern of results
supports the idea that LH function is necessary when prelexical phonology is required
to access the lexicon, while the RH can comprehend the words when access to the
lexicon is achieved by visual-orthographic codes, without phonological mediation.
Copyright 1990
Chisato Aoki
..
11
Dissertation Title: Hemispheric Lateralization of Japanese Kanji and Kana: Evidencefor Right Hemisphere Involvement in Semantic Processing ofKanji
Author: Chisato Aoki
NORTHEASTERN UNIVERSITY
Graduate School of Arts and Sciences
Department: Psychology
Approval for Thesis Requirements of the Doctor of Philosophy Degree:
.
\
I I^
Th sis Committee
11.//'.. , I. .{> I
,
^/
;I,
^^~
Head of Department
IC^.*.
I It fro,
I
^,. a
I ^ 9'0
.^ o
D to
I~zf'-?ODate
Dissertation Title: Hemispheric Lateralization of Japanese Kanji and Kana: Evidencefor Right Hemisphere Involvement in Semantic Processingof Kanji
Author: Chisato Aoki
Department: Psychology
NORTHEASTERN UNIVERSITY
Graduate School of Arts and Sciences
Approved for Thesis Requirements of the Doctor of Philosophy Degree:
,.
.
A^,
I^^I^;,,,
Th sis Committee
^.
Head ofIC
partment
Graduate School Notified of Acceptance:
A1"'Signed
^' ?'o
I
I^
I, 00
o
Copy Deposited in Library:
I- f"- 90
Signed
I 8 ?'Ote
I-^-^ODate
12 6z. -/,,-
te
I
,)
Date
help of many people. I would like to express my appreciation to my dissertation
adviser, Dr. Adam Reeves, who kindly allowed me to work independently, and gave me
his intellectual as well as technical support whenever I needed. I am especially
indebted to Dr. Adam Reeves for his tremendous patience and effort to read the
This dissertation could not have been completed without the contributions and
ACKNOWLEDGEMENTS
manuscript many times and improve my writing. I also would like to thank my former
adviser, Dr. Martin Block, for his guidance and support thoughout my graduate career.
I am grateful to the other members of my committee, Drs. Harold Goodglass, Nancy
Hildebrandt, and Nancy Soja, who gave me critical comments and constant
encouragement throughout my work.
I would like to thank anthe faculty members in the Department of Psychology,
who gave me an outstanding opportunity to learn about various fields of psychology in
the U. S. . I especially would like to express my gratitude to Dr. John Armington, who
advised me to stay in this program. Without his advice, I could not have completed my
Ph. D. in the U. S. . I also want to thank my fellow graduate students in the Vision Lab
and Language & Cognition Group, especially Marc Bearse, who acted as sounding
boards for my ideas.
. . .
111
My special thanks go to BTOhmsek Yen-ura, my wonderful assistant and very
good friend. He helped me with finding subjects, running experiments, inputing data
and correcting the manuscript of this thesis. More importantly, he cheered me up with
his unique jokes whenever I was down, and encouraged me throughout my work.
Without his help, I could not have completed my dissertation.
A special note of appreciation is extended to all of my Japanese subjects, who
kindly participated in my experiments. I would especially like to thank Mr. Sakamaki,
the head of Showa Women's Institute in Boston, who provided me with Japanese
students for my research.
encouraged me and supported me throughout my graduate career. Their faith and
confidence in me led me to the end of this long road.
Finally, I would like to thank my parents, thousands of miles away, who
IV
HEMISPHERIC LATERALIZATION OF JAPANESE KANJIAND KANA:EVIDENCE FOR RIGHT HEMISPHERE INVOLVEMENT
IN SEMANTIC PROCESSING OF KANJI
A Dissertation presented
by
Chisato Aoki
to
The Department of Psychology
In partial fulfillment of the requirementsfor the degree of
Doctor of Philosophy
in the field of
Psychology
Northeastern UniversityBoston, Massachusetts
January, 1990
.
,
Acknowledgements ................................................................................................... in
Chapter I : Introduction. ............................................................................................. I
Introduction ...................................................................................................... I
TABLE OF CONTENTS
Background ....................................................................................................... 4
I. The Nature of Kanji and Kana. ................................................................. 4
2. Cognitive Psychological Study .................................................................... 7
4. Tachistoscopic Visual Half-Field Studies ................................................ 12
5. Semantic Processingg in the Right Hemisphere ...................................... 15
Hypothesis ....................................................................................................... I 7
Chapter 2: Semantic Task with Kanji and Kana. .................................................. 18
3. Clinical Observations ...............................................................................----8
V
Introduction ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,..................................................... I ,
Experiment I: a short presentation of a lateralized picture. ....................... 20
Experiment 2: a long presentation of a lateralized picture. ........................ 26
Experiment 3: a long presentation of a centered picture. ........................... 30
Experiment 4: a long presentation of a centered picture. ........................... 33
Experiment 5: picture recognition ................................................................ 37
q
Chapter 3: Semantic and Phonological Processing of Kanji. ................................ 48
General Discussion ......................................................................................... 42
Introduction .................................................................................................... 48
Experiment 6: Picture-Kanji Matching with a centered picture ............... 50
Experiment 7: Picture-Kanji Matching with a lateralized picture. ............ 62
Experiment 8: Isolating the two tasks. .......................................................... 70
General Discussion ......................................................................................... 7 8
Chapter 4: General Discussion ................................................................................ 82
Summary of findings ..................................................................................... 82
Interpretations of the Finding. ...................................................................... 83
I. Semantic Processing ................................................................................... 83
2. Phonological Processing. .............................................,.............................. 88
3. Effect of Lateralizing the Picture ............................................................ 89
4. Condition Order Effect .................................................................------------89
5. Semantic Interference .....................................-----....------------""""""""""""
Directions for Future Research. .................................................................... 91
Tables ........................................................................................................................ 9 5
onclusion ...................................................................................................... 94
Figu res ...................................................................................................................... 9 7
References .............................................................................................................. I 28
Biographical Data ................................................................................................... I 36
Vl
Are different types of written scripts processed in different hemispheres? Does
the right hemisphere have some linguistic capacity? Japanese orthography is a good
tool with which to answer these questions because it has both syllabic characters, Kana,
and logographic characters, Kanji. These questions have gained considerable attention
in the last decade and are still being debated.
Chapter I: Introduction
usually recognized more accurately and faster in the right visual field (RVF) under
tachistoscopic presentation, while visuo-spatial stimuli, such as nonsense geometric
shapes and dots, are typically identified better in the left visual field (LVF) (e. g.
Benton, Hannay & Varney, 1975; Goodglass & Barton, 1963; Kimura, 1961, 1966, 1969;
Mishkin & Forgays, 1952). Because of the decussation of the optic tract, the RVF gives
direct access to the left hemisphere (LH), and the LVF to the right hemisphere (RH), so
these performance differences appear to point towards hemispheric specialization.
Because the two hemispheres communicate via the corpus callosum, direct evidence for
hemispheric specialization in normal subjects is difficult to obtain. Results from split
brain studies, however, do show clear-cut specialization (e. g. Gazzaniga, 1967, 1970;
Nebes, 1974). Based on the findings from alphabetic based studies, it seems reasonable
to hypothesize that alphabet-like characters, Kana, which represent vowels or
Introduction
It has been known from alphabetic-based studies that letters and words are
consonant-vowel pairs, are processed in a similarly way to phonetic or alphabetic
letters, while logographic characters, Kanji, which are pictorial and do not have fixed
sound values are possibly processed in a similar way to visuo-spatial stimuli.
The first studies of performance (e. g. recognition accuracy or latency) supported
the hypothesis that Kana (syllabic) characters are processed better in the RVF (thus the
LH) whereas Kanji(logographic) characters are processed better in the LVF (thus the
RH) (e. g. Hatta, 1976,1977, 1978, 1981; Hirata & Osaka, 1967; Sasanuma, Itoh, Mori &
Kobayashi, 1977). However, the performance advantage of Kanji can shift to the LH,
depending on the task required (Hatta, 1979,1981a, b; Hayashi and Hatta, 1982;
Sasanuma, Itoh, Kobayashi and Mori, 1980), the number of characters in a word (Hatta,
1978; Tzeng, Hung, Cotton and Warig, 1977), and concreteness (Elman et. a1. , 1983;
Hatta, 1977; 0hnishi & Hatta, 1980). As Paradis, Hagiwara and Hildebrandt (1985)
pointed out, these types of factors were not well controlled in some of these earlier
lateralization studies. Moreover, Kana and Kanji were rarely compared in a single
study.
Clinical studies of aphasia and alexia suggested that the RH might have some
linguistic capability. Japanese aphasics and alexics, who have damage to the LH, show
impairment in reading aloud both Kanji and Kana while their comprehension of Kanji
words is usually preserved (Sasanuma, 1975; Sasanuma, 1980, Sasanuma and Fujimura,
1972). English deep dyslexics, who usually have large LH damage, frequently fail to
read a written word aloud but can often comprehend the word either by gesture or
verbal circumlocution or by word-picture matching (Coltheart, 1980; Saffran & Mann,
1977; Patterson, 1981). Split brain studies also show that some of the patients can
comprehend the meaning of the words presented to their RHs (Zaide1, 1978; Zaidel and
Peters, 1981, but see Patterson and Besrier, 1984). These findings lead one to the
hypothesis that the RH, which does not know the phonology, has some low level
semantic capability (Coltheart, 1980, 1983).
The purpose of my research is to investigate if the two types of Japanese script
are processed more efficiently in different hemispheres of the human brain, and to
examine whether the RH has some capacity for semantic processing. More specifically,
I tested the hypothesis that the Japanese syllabic characters, Kana, are better processed
in the LH while the Japanese logographic characters, Kanji, are better processed in the
RH in the same semantic task. I also examined whether a task difference (i. e. a
semantic task vs. a phonological task) could make a difference to the hemispheric
2
processing of Kanji characters, I hypothesized that the relative dominance of the
hemisphere would change with the task given to the subjects.
To test the first hypothesis, I used a semantic task, namely a picture-word
matching task, for both Kanji and Kana characters under tachistoscopic conditions
using normal native Japanese people. In this task, pictures were matched to words
based on their meaning. To test the second hypothesis, I used the same picture-word
matching task in both a semantic and phonological form for Kanji characters. Subjects
matched meaning in the semantic task, and sound in the phonological task.
Consequently, there are two series of experiments in this thesis. The first series of the
experiments compared Kanji and Kana character processing in a semantic task
(Experiments I - 5 in Chapter 2). The second series compared the performance in
processing Kanji characters on two different tasks (Experiments 6 - 8 in Chapter 3).
In this chapter, I first describe the nature of Kanji and Kana characters' I then
review the studies suggesting differential processing of Kanji and Kana from the
cognitive psychological approach, from clinical observations of Japanese aphasia,
alexia, and partial split brain patients, and from tachistoscopic lateralization studies of
Kanji and Kana. I finally focus on the issue of RH language capability, and propose
the questions leading to the experiments described in Chapter 2 and Chapter 3.
I
3
I. The Nature of Kanji and Kana
Japanese orthography is unique since it has two types of written scripts. One
involves logographic or ideographic characters called "Kanji" which correspond to
morphemes, and the other involves phonetic characters called "Kana" which correspond
to mora or syllables. Kanjiis borrowed from Chinese characters which originated from
ancient hieroglyphics (Figure 1.1(a)). Only a small percentage of currently used Kanji,
however, retain high hieroglyphicity; most Kanji characters consist of multiple
elements and are far from pictographic. Japanese children start to learn the most
simple and highly hieroglyphic characters in the first grade, and usually learn 1850
Kanji characters in the 9 years of compulsory education. It is necessary to know
approximately 3000 Kanji characters in order to read a daily newspaper without
difficulty. However, most Japanese words consist of 2 or more combinations of Kanji,
so the number of words one must remember is far larger than 3000 (Sasanuma, 1980).
Kana is a Japanese invention derived from simplified Kanji characters to make
a one-to-one correspondence between a "mora" or syllable and a character (Fig. 1.1 (b)).
Kana is so simplified that it has lost the original shape of Kanji characters and does
not have meaning. There are two rather different ways of writing Kana, known as
Hiragana (cursive Kana) and Katakana (square Kana). Each type of Kana consists of
46 basic characters, making 71 symbols by adding diacritical marks like " and to the
basic characters, Since these n Kana characters represent all moraic units (consonant-
vowelsyllables), any word in spoken Japanese can be written in Kana. Kana is learned
formally in the first grade of primary school(by the age of 6).
These three types of scripts (Kanji and two Kana) are used in combination to
compose a sentence. Kanjiis used for content words to indicate meaning, such as
nouns and roots of verbs, adjectives and some adverbs. Hiragana is used for function
Background
4
words, the inflectional endings of verbs, adjectives and adverbs, some nouns, and
sometimes for "furigana" to show the pronunciation of difficult Kanji. Katakana is
restricted to representing loan words and some onomatopoetic expressions but
sometimes is used to emphasize a word or to command attention similar to the usage of
italics.
Although an entire sentence can be written only in Kana, as occasionally seen in
children's books, it would be difficult or even unintelligible to normal adult Japanese
readers, There are two possible reasons for this. One is that Kanji plays an important
role in building the semantic blocks to separate words so that the semantic unit as a
word stands out from the background Kana (there is no space between words as seen in
English. ) The other is to clarify the meaning of homophones. For instance, the sound
Ihashil has 3 meanings, bridge, edge and chopsticks. If this word is written in
Hiragana in the following sentence, it is impossible to distinguish which word it means.
""' I' L^.._C ^: ^.. IC,Watashi wa [hashi] o inita.I saw a [hashi].
However, if it is written in Kanji, one can immediately identify the meaning since
if'7^~ means a bridge, z;^ledge and I^'* means chopsticks.There are, in fact, a great number of homophones in Japanese. Iwata (1984), for
instance, listed 30 different Kanji characters for the sound I inl (see Figure 1.2).
Kanji have two to 64 homophones with the median of five (Zusetsu Nihong0, 1982).
"In a conversational situation, we (Japanese) often try to evoke the visual image of a
given homophone by asking the speaker which Kanji character corresponds to the
meaning of the word said. " (Sasanuma, 1980, p. 53)
There are three important attributes of Kanji as listed by Inoue, Saito and
Nomura (1979). First, Kanji has two basic readings, On-reading (Chinese reading) and
Kun-reading (Japanese reading). The On-reading is a pronunciation of a character that
was adapted from an original Chinese reading, while the Kun-reading is its Japanese
5
translation. Most of Kanji characters have at least one On-reading and one Kun-
reading, resulting in multiple readings of one character (Zusetsu Nihong0, 1982).
Second, most of Kanji consist of elements that carry semantic and/or phonological
information. In fact, sixty five percent of commonly used Kanji consist of both
phonological and semantic elements as shown in Figure 1.3 (Zusetsu Nihong0, 1982), the
pronunciations of the phonological elements usually being based on their On-readings.
The remainder have either a single element or elements without phonological
information. Therefore, when one encounters a new character, one may be able to
pronounce it by looking at the phonological part and also be able to make a close guess
at its meaning by looking at the semantic part. Third, a compound Kanjisometimes
has two different arrangements for its elements. For instance, 1:7 (mouth)
and I^~(now) can be arranged as either PI^'~(recite) or (^'~ (include) to make new words.
,
^ (thread) and 11El (field) can be arranged as either, ^!!zj(narrow) or ^^:, (accumulate).(Examples were taken from Inoue, Saito and Nomura, 1979). Thus, many Kanji
characters carry visual, semantic and phonological information.
These attributes of Kana and Kanjisuggest that the two types of orthography
carry different information: while Kana provides precise phonological information, but
less precise semantic information (less precise because of hornophonicity), Kanji
provides precise semantic information but less precise phonological information (less
precise because of multiple readings).
Based on the different linguistic functions and characteristics of the two scripts,
differential processing of the two scripts has been suggested from three different
methodological approaches: cognitive psychological, clinical, and tachistoscopic.
6
- -----------------
-- -------------------
Insert Figures 1.1(a)(b), 1.2 & 1.3
Japanese people normally claim that it is easier to comprehend the meaning of
a sentence if it is written in both Kanji and Kana, and that they can extract the
meaning of a Kanji word visually without pronouncing it. Cognitive psychologists
suggest that there might be a dualsystem, one involving sequential processing and the
other possibly parallel processing. They suggest that a Kana word is processed
sequentially in the order grapheme-phoneme-morpheme, while a Kanji word is
processed either sequentially but in the different order grapheme-morpheme-phoneme,
or in parallel from grapheme to morpheme and phoneme simultaneously; it is not clear
which processing takes place (Gory0, 1987: Inoue, Saito & Nomura, 1979).
To support differential processing for Kanji and Kana, it has been reported that
it takes more time to respond to Kanji words than to Kana words when subjects have
to read aloud, while the response latency is shorter for Kanjithan Kana when silently
reading the same words (Gory0, 1987; Sait0, 1981, but see Hirose, 1984,1985). This
comparison of Kanji and Kana might be unfair because some Kanji characters have
multiple pronunciations and involve more processing. However, Kana characters still
take less time to read aloud than Kanji characters that have only one Kun-reading
(Gory0, 1987).
2. Cognitive Psychological Studies
7
The other evidence of differential processing of the two types of characters
comes from concurrent vocal interference (articulatory suppression) experiments. It has
been reported that concurrent vocal activity, such as counting the numbers from I to 5,
interferes with reading in Kana but not in Kanji(Kimura, 1984). Kimura (1984)
interpreted this to mean that access to the lexicon for Kana words requires prelexical
phonology while access to the lexicon for Kanji words can be achieved without
phonological mediation.
Results from visual search experiments also support this idea. In a simple
search task, the latency for searching a target word embedded in background words
was shorter for Kana than for Kanji. However, when semantic processing was
involved, for instance, searching for a word which matched a previously given
category, it took less time to find Kanjithan Kana (Gory0, 1987). The results indicate
that a word written in Kanji has an advantage over a word written in Kana in a
semantic decision task. Gory0(1987) also found that short term memory for Kanji
words is less interfered with by auditory stimuli(listening to nonsense sentences) than
for Kana words. These results suggest that less phonetic coding may be required in the
semantic processing of Kanji words, supporting the idea of two different information
processing systems for Kanji and Kana.
Cognitive psychological studies generally support the idea that access to the
Kana lexicon requires phonological processing while access to the Kanjilexicon is
achieved via direct visual codes without phonological mediation (Gory0, 1987; Kimura,
1984; Sait0, 1981). However, the idea that Kana is processed only via phonological
codes has been challenged by recent research which has shown lexical access in Kana
also can be achieved without phonological processing when Kana words are highly
visually familiar (e. g. Besrier & Hildebrandt, 1987; Personal communication with
Hildebrandt, 1989; Hirose, 1984,1985).
The studies reviewed so far, while supporting differential processing of Kana
and Kanji, have usually not been concerned with the hemispheric localization of the
two types of scripts. The studies reviewed in the next section address more directly the
8
localization issue.
3. Clinical Observations
A number of clinical studies have reported that the majority of Japanese
aphasic and alexic patients show greater difficulty in processing Kana, while
maintaining relatively good preservation of Kanji(e. g. Sasanuma, 1974, 1975; Sasanuma
and Fujimura, 1972; Yamadori, 1975), although there are some rare cases in which
patients show a selective impairment of Kanji processing and a surprising preservation
of Kana processing (e. g. Imura, 1943, Saranuma and Monoi, 1975). Some patients with
global aphasia preserve an ability to match high frequency Kanji words to their
equivalent pictures.
Sasanuma and Fujimura's(1972) study in which they compared transcription
errors among aphasic patients with right hemisplegia and non-aphasic controls (both
right and left hemisplegics) supports the idea that Kanji and Kana involve different
information processing systems. They found that both the aphasic and non-aphasic
groups made errors in Kanjitranscriptions while only the aphasic group made errors in
Kana transcriptions. Furthermore, the overall error rate in the aphasics was worse in
Kana (55%) than in Kanji(38%). The types of errors made in Kanji and Kana among
aphasics were quite different. Phonological confusion accounted for 59% of the errors
in Kana while it did not occur in Kanji. Graphical confusions accounted for almost
50% of the errors in Kanji but only 0.4% of the errors in Kana. The majority of
graphical confusions (nearly 90%) consisted of the following two types: wrongly
compounding two or more units of a character (60%) and adding or omitting obligatory
strokes (30%). It is of interest that the types of errors made on Kanji and their
frequency distribution in both the aphasics and the non-aphasic controls followed the
same pattern. Their results indicate that the degree and type of impairment involved
in Kanji and Kana are quite different. While there were no differences in semantic
confusability, there was a difference between Kanji and Kana in graphic confusability.
This suggests a strong role of visual information in Kanji as compared to Kana.
Summaries of Japanese dyslexia with damage in the LH (Sasanuma, 1980, 1986:
Faradis at a1. , 1985) describe different types of impairment. A case of alexia with
aphasia of Broca's type (Kotani, 1935), and a case of alexia with agraphia (Sakamoto,
1940), showed superior performance in Kanji comprehension and severe impairment of
Kana. A patient with Gogi aphasia (a gross impairment of the lexical/semantic
processing) showed a selective impairment of Kanji processing and a remarkable
9
preservation of Kana performance. Pure alexia affects both Kana and Kanji,
producing a reading impairment accompanied by visual and semantic errors in Kanji
(Niki and Ueda, 1977), although Kanji comprehension is relatively well preserved
(Iwata, 1984; Iwata at a1. , 1982). For instance, the pure alexic patients often can
describe the object that is written in Kanji(e. g. arrow) and say to what it is related
(e. g. "This is what a warrior uses. "), or frequently can point to a picture to which the
word corresponds.
Based on observations of pure alexic patients who usually have a lesion in the
left medial occipital lobe, Iwata (1982) suggested a model of reading Kanji and Kana.
First, a visually represented word is processed in both left and right occipital lobes.
The visual information received in the right occipital lobe is transferred to the left
occipital lobe through the splenium of the corpus callosum. The information in the
left occipital lobe is then sent to the left angular gyrus which sends it, still in visual
form, to Wernicke's area, where the auditory memory of the word is evoked, and the
comprehension of the word is completed. In the case of pure alexia, since the left
occipital lobe is damaged, visual information presented in the RVF is not processed,
which causes the right homonymous hernianopia, typical of these patients. Information
presented in the LVF is sent to the left occipital lobe through the splenium of the
corpus callosum, but it is not processed further. Consequently, visual information
concerning the word is not converted to auditory information, so the word cannot be
read out. However, as was noted, alexic patients can show some semantic processing by
saying the word is related to something. So Iwata (1982) suggested that the
information arriving at the right occipital lobe is processed further in the RH to
comprehend the meaning of the word. Since the RH does not have any phonological
function, the patient is, however, unable to read aloud.
It seems that Kanji versus Kana reading dissociation depends on the location of
the brain damage and the strategy used for reading (Iwata at a1. , 1982).
Studies of partial split brain patients also support the idea of dissociation
10
between Kanji and Kana and support Iwata's model. It has been reported that severe
impairment of oral reading and comprehension of both Kanji and Kana occurs at an
early post-operative stage in the RH; however, two or three years later, a post-operative
retest showed recovery of Kanji but not Kana (Sugishita at a1. , 1978). A recent follow-
up study of a Japanese partial split brain patient at 9 to 12 years post-operation
showed that reading aloud or comprehension was much less disturbed for Kanjithan
for Kana in the LVF, while performance for both Kanji and Kana was equally good in
the RVF (Sugishita, at a1. , 1986; Sugishita and Yoshioka, 1987).
These results suggest that phonological processing is dominantly handled in the
LH, while some semantic processing might be handled. in the RH especially when
grapheme-morpheme processing is required, as may be the case in processing Kanji.
This is consistent with the results from split brain patients in the English language,
whose RHs show some comprehension of concrete nouns although they are unable to
match a printed word to a picture with a rhyming name or to a printed word that
rhymes, or two printed non-words that rhyme (See Coltheart, 1980; Zaidel & Peters,
1981). "The literature on brain-damage patients suggests that some processes such as
speech production, phonetic perception, and grapheme-phoneme conversion, are
exclusively localized to a single hemisphere whereas other processes, such as
comprehension of spoken or written nouns of high imagery value, may be handled by
both hemispheres though more efficiently by the left. " (MOScovitch, 1986, P. 96).
In summary, clinical observations suggest that processing Kanji and Kana
involves different locations of the brain. Kana processing usually involves the LH
while Kanji processing may involve both the left and RHs. When Kanji words must be
read aloud or processed phonologicalIy, the LH is also involved. If phonological
processing is not required, RH comprehension may suffice, although it is still not
known to what extend the RH has such a capacity. However, the results from the
brain-damaged patients may be obscure since the damage might involve areas other
than the specific locations we are interested in. Therefore, the results from normal
11
people using a tachistoscopic lateralization technique are helpful in examining this
issue. Such studies should also be helpful in assessing whether the dissociation shown
in brain-damaged patients reveals anything about the language processing of Kanji and
Kana in normals.
4. Tachistoscopic Visual Half-Field Studies
Tachistoscopic visual-half field research on written Japanese with normal
Japanese persons has suggested that there might be two different hemispheric systems
for processing the two types of scripts (Hatta, 1976, 1977a, 1978; Hirata & Osaka, 1967;
Sasanuma at a1. , 1977). The research shows a script/hemisphere interaction. Thus, in
simple recognition tasks or physical matching tasks, Kana is recognized more
accurately or faster in the RVF (the LH), while Kanjiis processed better in the LVF
(the RH). The results of Stroop tests have shown a greater interference effect for
Kanjiin the LVF than in the RVF. but no difference for Kana (Hatta, 1981b;
Monkawa, 1981). This is consistent with the idea that Kanji and visual information
are both processed in the RH while Kana are predominantly not RH.
12
originally thought. In a simple identification task, such as reporting single Kanji
characters verbally, or in a physical matching task using single Kanji characters, a
LVF advantage was indeed reported (Hatta, 1977, 1978, Hatta, 1981a). When Kanji was
presented as a compound word (2 character Kanji word), however, RVF superiority was
found (Hatta, 1978; Tzeng at a1. , 1979 for Chinese subjects). Tzeng at a1. (1979)
suggested that this result is consistent with the idea that the LH is better for sequential
and analytical tasks whereas the RH is better for holistic recognition tasks. Hatta
(1978), on the other hand, suggested that the LH more efficiently processes a two-
character Kanji word because it has only one pronunciation determined by the
combination of the two characters, whereas the RH more efficiently processes a single
However, the lateralization of written Japanese is not as simple as was
Kanji word because it has multiple pronunciations which might take more time for the
LH to process. This is in addition to the expected LH dominance in phonological tasks.
Indeed, reading Kanji phonologicalIy yields LH superiority (Sasanuma, at a1. , 1980).
When subjects were asked to decide if two words had the same pronunciation in Kanji,
or rhymed in Kana, their responses were more accurate in the RVF for both Kanji and
Kana, but more asymmetry was found for Kana than for Kanji.
Hatta (1979, 1981a) has suggested that Kanji processing involves LH functioning
when higher level processing is required. The results of his research show RH
superiority in simple physical recognition of single Kanji, no significant difference in
a lexical decision task, and a LH superiority in a semantic congruence task. These
outcomes suggest that there is a progressive shift towards the use of the LH in Kanji
processing as the task moves from simple visual identification to making linguistic
decisions. Hatta (1981) and Hayashi and Hatta(1982) also obtained a RVF superiority
in a semantic categorization task using Kanji, suggesting that semantic processing
occurs in the LH.
The superiority of the RH in simple recognition tasks may be the result of
treating Kanji as whole figures (Tzeng at a1. , 1979), which consequently require
processing by the RH. The lexical decision task may require finer attention to detail,
so reducing the role of the RH. The semantic congruence task seems to require greater
LH functioning, since a Kanji character is not treated simply as a visual input, but is
read with a particular pronunciation and meaning.
These results suggest that it may not merely be a property of scripts that
determines the direction of lateralization, but rather the nature of the task
requirements (phonological, visual matching or semantic) (Sasanuma at a1. , 1980) or the
level of processing required to perform the task (speech or visual'recognition, lexical or
semantic decision task) (Hatta, 1981a).
Summarizing a large literature, five hypotheses have been advanced to account
for hemispheric differences in processing Kanji and Kana characters' The first
13
hypothesis is a "localization of scripts" hypothesis. That is, Kana is localized in the LH
while Kanjiis localized in the RH, regardless of any other factors. This hypothesis,
however, seems to be unlikely because some studies have found a shift of hemispheric
advantage depending on task demands (e. g. Sasanuma at a1. , 1980; Hatta, 1981b; Hatta &
Hayashi, 1982), and depending on levels of processing (e. g. Hatta, 1981a).
The second is a "localization of function" hypothesis. It has been suggested that
linguistic processing, especially phonological processing, takes place in the LH (e. g.
Zaidel and Peters, 1981), while visuo-spatial processing takes place in the RH (e. g.
Kimura, 1966). According to this hypothesis, Kana should be lateralized in the LH
' since reading Kana characters usually requires phonological processing. In contrast,
Kanji should be lateralized in the RH because Kanji characters contain complex visual
information and phonological processing might not be necessary to access to the lexicon
for Kanji. Sasanuma at al. 's (1980) findings seem to fit this hypothesis nicely.
The third is a "level of processing" hypothesis. According to this hypothesis, a
low level of processing, such as phonological and semantic processing takes place in the
RH while more higher level of processing, such as linguistic processing, takes place in
the LH (e. g. Cohen, 1972; Hatta, 1979, 1981a). However, this level of processing
hypothesis might be a subset of a localization of function hypothesis since a lower
level of processing mentioned here (i. e. physical matching) corresponds to visual spatial
processing while a higher level of processing corresponds to linguistic processing.
The fourth is a "sequential-parallel" hypothesis. According to this hypothesis,
Kana is processed sequentially, or analyticalIy, thus in the LH, while Kanjiis
processed parallelly or holistically, thus in the RH. This hypothesis seems to work for
single character Kanji, but cannot explain the LH advantage for two-character Kanji.
The fifth is a "competition between right and LH-governed components"
hypothesis (See Goodglass and Butters, 1988). According to this hypothesis, the same
stimulus may be processed mainly in the RH when a visual perceptual decision is
necessary and primarily in the LH when a verbal or categorical decision is necessary.
14
The findings of Hatta (1979, 1981a) might also fit in this hypothesis, although
materials used in Hatta's studies were different characters in different experiments.
The second and fifth hypothesis are relevant in the current study as can be seen
below. Another hypothesis, which does not directly address the issue of lateralization
of Kanji and Kana but is relevant in this thesis, is a "hemispheric activation"
hypothesis first proposed by Kinsbournne (e. g. 1970, 1973). He suggested that prior
activation of a hemisphere enhances the efficiency of the hemisphere, so that stimuli
which are normally processed equally in both hemispheres will be processed better in
the already activated hemisphere. Stimuli which are processed better in one
hemisphere will be processed even better in that hemisphere if it is already activated.
However, if both hemispheres are equally activated well, then such an advantage might
be reduced or might disappear.
5. Semantic Processing jin the Right Hemisphere
Although research with normal Japanese subjects suggests that the semantic
processing of Kanjitakes place in the LH, this is not fully established. Firstly, the
experimental evidence is confounded because the parameters of the stimuli in early
studies were not well controlled, as mentioned earlier (Faradis at a1. , 1985). Secondly,
the clinical evidence suggests a possible involvement of the RH in comprehension of
meaning of Kanji words. Thirdly, the semantic tasks used in tachistoscopic studies
require more than the comprehension of meaning. In the previous semantic tasks, the
subjects were required to make a judgment about the spatial position of the word or
the superordinate category of the word as well as determine its meaning. It may be
that the RH would be better at a simple semantic task, such as the picture-word
matching task used in the clinical research.
15
lones and Aoki(1988) tested this possibility. The task used in their study was
to judge if a picture and a word matched. If Kanji processing does not require
phonological processing and the visual representation of Kanji evokes the meaning of
the concept, the RH might process this task more efficiently than the LH, or at least
both the right and LH might have the same efficiency. The stimuli were 40 highly
concrete, highly hieroglyphic and highly familiar single Kanji words selected from
Kitao at al. 's (1977) list of 881 Kanji, and 40 equivalent Kana words. Half the words
were matched with the pictures and the other half were not.
The results indicate that the threshold stimulus duration (the duration required
for picture-word matching at an accuracy of 75 %) is lower in the LVF than in the
RVF for Kanji words, while the threshold for Kana words is lower in the RVF than in
the LVF. This suggests that the RH processes the highly concrete, highly hieroglyphic
and highly familiar single Kanji words more efficiently, even in a semantic task, while
Kana is, as usual, processed better in the LH. In contrast to the earlier research, the
nature of the task (picture-word matching) did not require overt phonological
processing (visual-auditory association) of Kanji words and did not require
categorization. Kana may still have been read phonologicalIy even though this was not
required.
However, there were some uncontrolled factors in the experiment. Half the
Kanji words were not matched with the pictures. These consisted of three types:
16
mismatched real characters that resembled the correct real characters, modified
characters whose radicals were replaced with other radicals, and nonsense characters
whose strokes were either added to or omitted from real characters, It is not clear if
subjects performed the task based on semantic decision, lexical decision or visual
discrimination, because the method did not distinguish among errors made in response
to these three types of mis-match. Therefore, it is necessary to examine if a RH
advantage would be still obtained in a semantic task when the visual similarity of
Kanji words is reduced, and no lexical decision is required.
It is also interesting to investigate whether measurements made on above
threshold stimuli would also show the interaction obtained with the threshold
measurements in Jones and Aoki's (1988) study. They found that in order to achieve
equal accuracy, it was necessary to use quite different stimulus durations in the two
hemispheres for the two types of script. However, when accuracy was experimentalIy
equated, reaction times were also equal. In this thesis, the main dependent variable
was reaction time while stimulus duration was fixed. Error rates were also analyzed as
performance was not perfect. As the stimuli were above threshold, the ability of the
visualsystem to encode them was generally not in doubt, so errors are unlikely to have
a purely sensory origin
I tested two hypotheses; first, that the RH would be better than the LH in
processing the meaning of Kanji words, but the opposite pattern would hold for Kana
words. Second, that, the relative dominance of the hemispheres for Kanji would
change with the task given to the subjects. The first hypothesis was tested in
Experiments I-5 in chapter 2. In these experiments, pictures were matched to words
based on their meaning (the semantic task). Reaction times were measured. A cross-
Hypotheses of the Current Study
over interaction between character type and visual field (thus hemisphere) was
17
expected. The second hypothesis was examined using only Kanji words in Experiment
6-8 in chapter 3. The Kanji words were tested not only in the semantic task, but also
in a phonological task, in which the sound of the words was matched to the sound of
the picture's name. I again expected an interaction between task and visual field.
In this chapter, I examine whether semantic processing of Kanji and Kana
involves opposite hemispheres. As described in Chapter I, semantic processing of Kanji
has been studied in higher-level semantic tasks (Hatta, 1979, 1981a; Hayashi & Hatta,
1982), whereas semantic processing of Kana has been rarely studied. Since the same
lexicon can be represented by both Kanji and Kana, it is interesting to investigate
whether access to the lexicon represented by either Kanji or Kana more dominantly
involves one hemisphere than the other.
Chapter 2: Semantic Task with Kanji and Kana
Introduction
Alithe Kanji characters used in this study were highly hieroglyphic according
to Japanese norms studied by Kitao at a1. (1977). They were also highly concrete and
highly familiar, which previous research suggests should favor the RH (e. g. Day, 1977;
Elan at a1. , 1981; Ellis & Shepherd, 1974; Hines, 1976; 0htani and Hatta, 1980). These
Kanji characters tend to be more pictorial and consequently should be more efficiently
processed in the RH. Such increased efficiency should cause a hemispheric shift in
performance on tasks in which these factors occur either randomly or in a less pictorial
combination as in previous studies (e. g. Hatta, 1979, 1981a; Hayashi & Hatta, 1982;
Sasanuma at a1. , 1980). Therefore, the present study was designed to examine if high
pictorialness will produce a RH superiority in the semantic processing of Kanji.
In this chapter, I describe four experiments comparing processing of Kanji and
Kana in a picture-word matching task. In these experiments, stimulus durations were
fixed, and reaction times and error rates were measured. I predicted that Kanji
characters would be relatively favored in the RH while Kana characters would be
18
relatively favored in the LH.
side of the visual field, which is the same presentation condition as in lones and Aoki
In Experiment I and 2, both a picture and a word were lateralized in the same
(1988). I expected a cross-over interaction between character type and visual field.
The picture presentation durations, however, were different between the two
experiments. This manipulation was performed to examine whether the clarity of the
picture plays a role in lateralization, as suggested by Sergent (1982, 1986).
I then presented a picture in the center of a screen, so that both hemispheres
could process it equally, and be equally activated by it. Picture duration was short in
Experiment 3 and long in Experiment 4. These two experiments were conducted, in
contrast to the first two experiments, to investigate the possible role of prior
hemispheric activation on lateralization of the two types of script. Relevant here is
the "hemispheric activation" theory of Kirisbournne (e. g. 1970, 1973), desc, ibed in
Chapter I, which suggests that a lateralized picture presentation would maximize the
amount of interaction between character type and visual field. For example, when a
picture is presented to the LVF and thus enhances the activity of the RH, an incoming
Kanji word would gain advantage because Kanjiis usually processed better in the RH.
However, a Kana word would not gain any advantage, because Kana needs to be
processed in the LH, and activation of the RH would not enhance the processing
efficiency of Kana. On the other hand, when a picture is presented to the RVF and
thus enhances LH activity, the Kana word would gain an advantage, but the Kanji
word would not. When a picture is presented in the center of the visual field, the
picture activates both hemisphere equally. Thus, the asymmetry might be suppressed
by overall activity of the two hemispheres, and may not be observed. Therefore, the
character-visual field interaction might be reduced or might even disappear. I thus
19
expected a larger interaction in Experiments I and 2 and a smaller interaction in
Experiments 3 and 4.
In Experiment 5, I examined if picture processing requires a different input
duration in different hemispheres to reach the same performance level(i. e. correct
responses), as a control for the previous Experiments.
In this experiment, both a picture and a word were presented to the same visual
field successively for a short period. I expect an interaction between character type
and visual field; Kanji being relatively faster in the LVF (RH), and Kana being
relatively faster in the RVF (LH), in a simple semantic task.
Experiment I: a picture-word matching task
with a short presentation of a lateralized picture
Subjects: Subjects were twelve Japanese people (4 males and 8 females), with
mean age of 25, who regularly read and spoke Japanese. They had resided in Boston
for, on average, one and half years, ranging from I to 54 months. (The numbers of
male and female subjects were not equal due to limited availability of subjects. )
According to the standards of the simplified Edinburgh Handedness Inventory (see
Bryden, 1977; Oldfield, 1971), all of the subjects were right handed and none of their
immediate family members were left handed. All of the subjects had normal or
corrected to normal vision (at least 20120). The subjects were paid $6.00 per hour to
participate in the experiment.
Method
20
Apparatus: A Gerbrand GII71 two-channel projection tachistoscope, which
consisted of two Kodak Ektagraphic projectors and electrical shutters with a 4-channel
digital millisecond timer (Gerbrands "300-C" series), was used to project stimuli to a
translucent rear projected screen set 66 cm away from the subject at eye level. The
projection field subtended 24 degrees of visual angle horizontally and 16 degrees of
visual angle vertically. The field was white during a stimulus flash. Otherwise, the
field was dark.
The reaction time was measured by Hunter Klockcounter (Model120 A series D)
to the nearest millisecond from the onset of the stimulus presentation controlled by
the tachistoscope to the subject's key press. Two microswitches, connected to the
Klockcounter, were mounted on a board and served as the response keys. A meter
indicated which key was pressed.
Stimuli: There were 40 Kanji words, 40 Kana words, and 40 pictures. Anthe
Kanji words were single characters, The Kana words were transcriptions of the Kanji,
and the pictures depicted the Kanji. Highly pictorial Kanji characters were selected
using Kitao at al. 's list (1977) of 881 Basic Kanji characters which are high on
concreteness (the mean score of 90.7 out of 100), hieroglyphicity (the mean score of
4.59 out of 7) and familiarity (the mean score of 4.72 out of 7).
Each stimulus was hand-written with black ink in a two centimeter square on a
white paper. For Kana, each character was written in each square. Then, the stimuli
were individually photographed with black and white film under constant illumination
so that the intensities of alithe stimuli were the same. The pictures and Kanji words
were 1.5 degrees of visual angle in width and height. Kana words were 1.5 degrees of
visual angle in width and 3.5 degrees of visual angle in height. A small digit, 0.4
degrees of visual angle, was placed at the center of the picture or the word.
The stimuli were rear projected onto the screen. Each stimulus appeared at 5
degrees of visual angle from the central fixation point on the left or right depending
on the condition. The luminance of the background was 41 ft. -L. , and the contrast of
the stimulus to the background was 94% (the character luminance matched the field
seen through a 1.28 log unit filter).
21
Procedure: The experiment consisted of two sessions: a picture-Kanji matching
task and a picture-Kana matching task. Each subject was instructed to sit at a table
facing the translucent screen with his/her head on a chin-rest to minimize head
movement. Prior to the experiment, the subject saw all 40 pictures and correctly
identified them. This was necessary in order to make sure that the subject understood
all the pictures.
seconds, followed by the presentation of a picture for 150 msec, then by an
Each trial consisted of sequence in which a fixation point was presented for 2
interstimulus interval of I second, after which a word was presented for 150 msec.
The subjects viewed the stimuli binocularly.
On half the trials, the fixation point was replaced by a cross when the picture
was presented and by a digit (from 2 to 9) when the word was presented. On the
remaining trail, the cross and digit were presented in the reverse order.
matched or mismatched by pressing the appropriate response key as fast and accurately
The subject's task was to indicate whether the picture and the word were
as possible. (Half the time, they did match, and half the time, they did not. The
matched and mismatched pairs were randomly selected. ) These responses were made by
the index and middle finger of the right hand. The use of fingers for yes and no was
counterbalanced across the subjects. The subjects then reported the fixation digit. If
the subject incorrectly reported the digit being presented, then the results for that trial
were omitted on the assumption that the subject was not fixating. This happened in less
than 2% of the trials, in the subsequent experiments.
22
Subjects had 20 practice trials to familiarize them with the task prior to testing.
This practice run was followed by testing either picture-Kanji or picture-Kana
sequence. After the first block of trials, subjects were given a 10 minute break and
then tested for the other sequence following new 20 practice trials. The experimenter
recorded reaction time, and whether the response and the reported fixation digit were
correct or incorrect. The experiment took approximately one hour,
Design: The 40 picture-word pairs were presented once in each visual field,
resulting in 80 trials in one block. Kanji and Kana words were tested in separate
blocks. The order of presentation was counterbalanced across the subjects.
Reaction time: Reaction times for correct responses were used in the following
analysis. These reaction times are transformed logarithmically prior to the analysis.
This made the distributions approximately normal and homogeneous. The means of
these reaction times were then transformed back to reaction times in milliseconds by
taking antilogarithms. Consequently, the means indicated in tables and graphs are
geometric instead of arithmetic means.
An ANOVA with two within-subject factors was performed on the log reaction
times in correct trials. The two factors were visual field (LVF vs. RVF) and character
type (Kanji vs. Kana). The ANOVA showed no main effect of visual field (F(I, ll) =
1.28, p > 0.05), and 00 main effect of character type (F(I, ll) = 2.31, p > 0.05). The
analysis, however, showed a significant interaction between visual field and character
type (F(I, ll) = 20.3, p < 0,001). A summary of the results are shown in Figure 2.1(a)
and Table 2.1(a).
Results
23
(841 msec and 894 msec, respectively, p < 0.01) while Kanji was faster in the LVF thanRVT'
in the r'sl, t (811 msec and 831 msec, respectively, p < 0.05 with one tailed test). A
posthoc comparison with Newman Keuls Test showed that Kanji was faster than Kana
in the LVF (811 msec and 894 msec, respectively, p < 0.01) although there was no
significant difference between Kanji and Kana in the RVF (832 msec and 841 msec,
respectively, p > 0.05). The post comparison test was used for these comparison because
L. Vt'Planned comparisons showed that Kana was faster in the RVF than in the *e^
------------------------------------------
Insert Figure 2.1(a) and Table 2.1(a)
,
these results were not predicted from the previous studies.
Figure 2.2(a) shows individual data. The abscissa represents the difference
between the RVF and the LVF for Kanji, while the ordinate represents that for Kana.
A positive number indicates a RH advantage and a negative number indicates a LH
advantage. If we take an arbitrary 20 msec difference between the two visual fields as
a criterion for hemispheric advantage, the following observations are obtained from
the figure. For Kanji, 6 subjects showed a RH advantage, 4 subjects showed no
hemispheric difference, and only 2 subjects showed a LH advantage. On the other
hand, for Kana, 9 subjects showed a LH advantage, 2 subjects showed no difference
and only I subject showed a RH advantage. The subject who showed a RH advantage
for Kana also showed a strong RH advantage, which indicates that this subject is, for
some reason, RH oriented. The individual results, in general, support the overall
tendency that Kana is processed better in the LH while Kanjiis processed better in the
RH.
factors of visual field and character type showed that there was no main effect of
character, 3.5% for both characters (F(I, ll) < I), but there was a main effect of visual
field, 4.2% for RVF and 2.8% for LVF (F(I, ll) = 6.44, p < 0.5). However, there was 00
interaction between character type and visual field (F(I, ll) < I). (See Figure 2.3(a) and
Table 2.2(a).) Although there were fewer errors in the LVF than in the RVF, the
difference was very small(1.4%). This difference might be due to picture recognition
rather than character recognition since some subjects said that some pictures were hard
to identify.
Error rate: The overall error rate was about 3.5%. An ANOVA with two within
24
Insert Figure 2.2(a)
Insert Figure 2.3(a) and Table 2.2(a)
visual field and the character type in error rates. Consequently, the discussion,
There is no speed-accuracy-trade-off since there was no interaction between the
hereafter, is based on the reaction time results.
I found that reaction time is an indicator of laterality between Kanji and Kana.
The results suggest that the performance levels of Kanji and Kana in the semantic task
are lateralized when the stimulus durations in both visual field are equal. This extends
lones and Aoki(1988), who found the same pattern of interaction when stimulus
Discussion
duration was varied to obtain constant (75%) accuracy. Kanji characters are processed
faster in the RH while Kana characters are processed faster in the LH. Moreover, the
magnitude of difference between the two hemispheres was larger in Kana than in
Kanji. That is, Kana showed a clearly significant difference, but Kanjishowed a
marginal difference in Experiment I. This might mean that Kana processing is more
lateralized in the LH, but Kanji processing is bilateralized.
These results are consistent with earlier work on physical identification of
Kanji, which showed a RH advantage, but are inconsistent with the results from the
previous semantic tasks (e. g. Hatta, 1981a; Hayashi and Hatta, 1982), which showed a
LH advantage for Kanji. There is no literature which tested Kana laterality in
25
I
semantic processing, however. It is important to use both Kanji and Kana for the same
task for the same individual because we can reduce individual vanability and examine
processing difference between Kanji and Kana within the individual. The data in this
study show a different processing pattern for Kanji and Kana words which represent
the same meaning.
characters are processed dominantly in the LH. Kana words used, however, were
Previous researchers have focused on Kanji in their studies assuming that Kana
mostly transcribed from Kanji words. Therefore, Kana words were less familiar than
Kanji words. Besrier and Hildebrandt (1987) found that familiar Kana words were
named faster than unfamiliar Kana words (i. e. , Kana transcriptions from Kanji words),
which, they argue, means that lexical access to familiar Kana words may be achieved
without phonological mediation. A frequency effect was also obtained in low-
frequency irregular English words (Waters & Seidenberg, 1985), supporting Besrier and
Hildebrandt's argument. If pre-lexical phonology is not necessary for familiar Kana
words, use of these words may result in a RH advantage. This hypothesis is yet to be
tested.
In Exp. I, some subjects reported that pictures were hard to identify due to brief
presentation (exposure time of 150 msec). This difficulty in identifying pictures may
interfere with the picture-word matching task since a word is presented while a subject
is stintrying to identify the picture. This might cause a biased RH activation due to
the RH's superiority for identifying geometric figures, and might account for the
different in error rates between the visual fields. In order to eliminate this possibility,
I used a longer picture presentation duration in Exp. 2 so that all the pictures were
clearly identified.
26
This experiment was designed to examine the effect of picture duration on
Kanji and Kana lateralization. The picture was presented for longer period of time in
this experiment to clarify the picture identity. Presenting a picture for longer period
should reduce any possible extra involvement of the RH due to ambiguous picture
Experiment 2: a picture-word matching task with
a long presentation of a lateralized picture
presentation, and might eliminate the visual field difference in error rates in
Experiment I.
Subjects: Twelve new right handed Japanese people (3 males, 9 females), who
resided in Boston for, on average, I year (ranging from I to 48 months), served as
subjects. The mean age of the subjects was 27. The criterion for selecting subjects was
the same as in Exp. I.
Apparatus: The apparatus was the same as in Experiment I.
Method
The procedure was almost the same as in Exp. I except that the picture stimulus
Stimulus and Procedure: The stimuli were the same as those used in Exp. I.
duration was 500 msec. The picture-word pairs were presented either in the LVF or
RVF, but always in the same visual field, as in Exp. I. Since the picture duration was
more than 200 msec, there was a possibility that the subject moved his/her eyes to the
picture. In order to reduce this possibility, throughout the experiment, the subject was
reminded to fixate on the central fixation and not on the stimuli. The fixation digits
were also presented to control for the eye movement as in Exp I. If the subject
reported the number incorrectly, that trial was omitted from the analysis. This
27
occurred only 2% of the entire trials.
The results were essentially the same as the results obtained in Exp. I, although
both Kanji and Kana showed strong significant differences between the visual fields.
Reaction time: The reaction times were logarithmically transformed prior to the
analysis to make a distribution approximately normal and homogeneous as in Exp I.
An ANOVA with two within-subject factors was performed on the correct responses.
Results
The two factors were visual field (LVF vs. RVF) and character type (Kanji vs. Kana).
The ANOVA showed 00 main effect of character type (F(I, ll) < I), and 00
main effect of visual field (F(I, ll) = 3.69, p > 0.05). The analysis, however, showed a
significant interaction between visual field and character type (F(I, ll) = 5.96, p < 0.05).
A summary of the results are shown in Figure 2.1(b) and Table 2.1(b).
(908 msec and 940 msec, respectively, p < 0.01) while Kanji was faster in the LVF
than the RVF (827 msec and 863 msec, respectively, p < 0.01). The directions of
laterality for Kanji and Kana are the same as the previous experiment. Therefore, the
LVF-RH advantage for Kanji obtained in the previous experiment was not due to extra
involvement of the RH caused by ambiguous picture processing.
Planned comparisons showed that Kana was faster in the RVF than the LVF
Insert Figure 2.1(b) and Table 2.1(b)
subjects showed no difference, and only one subject showed a LH advantage for Kanji.
On the other hand, the figure indicates a LH advantage in seven subjects, no
Figure 2.2(b) indicates that seven subjects showed a RH advantage, seven
difference in one subject, and a RH advantage in four subjects for Kana.
28
InterestingIy, the four subjects who showed a RH advantage for Kana also showed a
strong RH advantage for Kanji, as seen in Exp. I. Similarly, the subject who showed a
LH advantage and two subjects who showed no difference showed a strong LH
advantage for Kana. Therefore, some subjects are more LH oriented and some others
are more RH oriented. However, the overallresults show a RH advantage for Kanji
and a LH advantage for Kana.
Insert Figure 2.2(b)
within factors of visual field and character type showed that there was no main effect
of character type, 2.1% for Kanji and 3.1% for Kana (F(I, ll) = 2.53, p > 0.05), 00 main
effect of visual field, 2.5% for RVF and 2.8% for LVF (F(I, ll) < I, p > 0.05). The
ANOVA also showed no significant interaction between character type and visual field
(F(I, ll) < I, p > 0.05). (See Figure 2.3(b) and Table 2.2(b).) The smaller error in the
LVF obtained in Exp. I disappeared when the picture was clearly presented in Exp. 2.
Therefore, the visual field difference obtained in Exp. I might have resulted from
difficulty in identifying a few of the pictures.
Error rates: The overall error rate was about 2.7%. An ANOVA with two
Insert Figure 2.3(b) and Table 2.2(b)
speed-accuracy trade-off. Therefore, the discussion is based on the reaction time
Since the error rates did not show any significant difference, there was no
results, as for Experiment I.
The reaction time results indicate that Kanji words were responded to faster in
the RH than in the LH, while the opposite occurred in Kana words. Because the
29
picture presentation was longer, the interaction between the visual field and the
character type likely resulted from character processing, not from picture processing as
might have occurred in Experiment I.
The results from the Experiment I and 2 strongly support the hypotheses that
Kanji and Kana are processed better in opposite hemispheres, and that the RH has a
semantic capacity especially in the case of processing logographic characters, such as
Kanji. The two experiments, however, lateralized both pictures and words as input
Discussion
stimuli.
The next two experiments were designed so that the picture presentation would
activate both hemispheres equally prior to word presentation. According to
Kmsbourne's activation hypothesis described earlier, presenting a picture in the center
of the visual field should reduce the lateralization effect for both Kanji and Kana
because activating both hemispheres prior to word recognition would make them both
ready to process any incoming stimuli. Consequently, the character-visual field
interaction would be reduced or eliminated.
This experiment examines if the same laterality effect shown in the previous
two experiments is obtained when the picture is presented in the center of the visual
field. For comparison with Experiments I and 2, Experiment 3 used a short picture
duration while Experiment 4 used a long duration.
Experiment 3: A picture-word matching task
with a short presentation of a centered picture
30
Subject: Twelve Japanese women (mean age of 19.3) who resided in U. S. for
about 2 months served as subjects. The criterion for selecting subjects was the same as
in Exp. I.
Apparatus: The apparatus was the same as the previous experiments.
Method
Pictures and words were presented for 150 msec, as in Experiment I. The procedure
was the same as in Exp. I, except that pictures were presented in the center of the
visual field. (As in Experiment I, fixation digits were used to prevent eye movement,
Stimuli and Procedure: The stimuli were the same as those used in Exp. I.
except that these digits were not only shown with the words. )
Reaction time: The reaction times were logarithmically transformed prior to the
analysis as in Exp. I. An ANOVA with two within factors for correct responses was
performed. The two factors were visual field (LVF vs. RVF) and character type
(Kanji vs. Kana).
The ANOVA showed 00 main effect of character type (F(I, ll) < I, p > 0.05),
and no main effect of visual field (F(I, ll) =3.69, p > 0.05). The analysis, however,
showed a significant interaction between visual field and character type (F(I, ll) =
5.96, p < 0.05). A summary of the results are shown in Figure 2.1(c) and Table 2.1(c).
Results
Planned comparisons showed that Kana words were responded to faster in the
RVF than the LVF (861 msec and 906 msec, respectively, p < 0.01) while Kanji did not
show any significant difference between the visual fields (879 msec and 877 msec,
respectively, p > 0.05).
Figure 2.2(c) indicates that, for Kanji, four subjects showed a RH advantage,
four subjects showed no difference, and four subjects showed a LH advantage,
resulting no overall hemispheric differences. On the other hand, for Kana, 11 subjects
showed a LH advantage and only one subject showed a RH advantage. Therefore, the
results revealed that Kana is processed better in the LH for the majority of the
subjects, whereas Kanjiis processed equally in the two hemispheres perhaps with some
31
Insert Figure 2.1(c) and Table 2.1(c)
individual differences.
factors of visual field and character type showed that there was no main effect of
character type (1.3% for Kanji and 1.6% for Kana, F(I, ll) < I) and no main effect of
visual field (1.2% for RVF and 1.6% for LVF, F(I, ll) < I). The ANOVA also showed
no significant interaction between character type and visual field (F(I, ll) < I). (See
Figure 2.3(c) and Table 2.2(c).) The LVF advantage in error rates found in Exp. I
disappeared in this experiment as well as in Exp. 2. This also supports the hypothesis
that the LVF advantage observed in error rates in Exp. I was due to ambiguity of the
pictures since the presentation of the picture in the center of the visual field made the
Error rate: The overallerror rate was about 1.4%. An ANOVA with two within
I
Insert Figure 2.2(c)
picture identification better (although the duration of the picture presentation was still
150 msec).
32
Since the error rates did not show any significant difference while the reaction
time showed a significant interaction, there was no speed-accuracy trade-off.
Therefore, the following discussion is based on the reaction time results.
Insert Figure 2.3(c) and Table 2.2(c)
the LH advantage was somewhat reduced for Kana, relative to when the picture was on
the side; and the RH advantage for Kanji disappeared altogether. This is in line with
the hypothesis that central presentation of the picture reduces hemispheric differences.
However, the reduction of the LH advantage for Kana, from 33 msec to 45
The results of Experiments I and 3 show that when the picture was centered,
Discussion
msec, was small and not significant. Therefore, alternative hypotheses should be
considered. When a picture is presented in the center of the visual field, it activates
both hemispheres. Since Kanji characters can be processed in both hemispheres, and
the hemispheres are equally primed by prior visualstimuli, both hemispheres are ready
to process Kanji characters when they subsequently appear. Furthermore, as shown in
the individual data (Figure 2.2(c)) each individual might have used a different strategy
Consequently, the averaged performance of the left and RHs are the same.
On the other hand, Kana characters still need to be processed in the LH due to
the phonological nature of character processing. Since the RH does not have a
phonological processing capability, activating the RH does not enhance response times
for Kana characters, Therefore, the LH stillshows an advantage over the RH in the
case of Kana.
Another interpretation is that a brief presentation of the picture might not have
activated both hemispheres fully, so that strong phonological processing due to visually
unfamiliar Kana words might have been manifested. If the picture duration is longer,
then both hemispheres would be fully activated and both Kanji and Kana laterality
effect would be reduced or eliminated.
33
the center would affect the laterality effect of both Kanji and Kana.
This experiment was designed to examine whether a long exposure of picture in
Experiment 4: a picture-word matching task
with a long presentation of a centered picture
representation of a picture should be clearer when it is presented in the fovea for one
second (this experiment) than when presented in fovea for 150 msec (Exp. 3), or when
presented peripheralIy for 500 msec (Exp. 2). If both hemispheres are fully activated
by a clear picture presentation, they should be ready to process both Kanji and Kana
equally. Therefore, the character-visual field interaction would be reduced even
The
further in this experiment.
Subject: Twelve Japanese women (with a mean age of 24.6), who had resided in
the U. S. for about 2 years (ranging from I to 60 months), served as subjects. The
criterion for selecting subjects was the same as in the previous experiments.
Apparatus: The apparatus was the same as the previous experiments.
Stimulus and Procedure: The stimuli were the same as those used in the
previous experiments. The procedure was almost the same as in Exp. 3 except that
picture duration was I sec. The picture was presented in the center of the visual field
as in Exp. 3.
Method
Reaction time: An ANOVA with two within-subject factors was performed on
log reaction times for correct responses. The two factors were visual field (LVF vs.
RVF) and character type (Kanji vs. Kana).
The ANOVA showed no main effect of character type (F(I, ll) < I), and no
main effect of visual field (F(I, ll) = 1.90, p > 0.05). It, however, showed a significant
interaction between visual field and character type (F(I, ll) = 4.92, p < 0.05). A
summary of the results are shown in Figure 2.1(d) and Table(d).
34
Results
----------------------------------------
Planned comparisons indicated that the LVF was faster than the RVF for Kanji
Insert Figure 2.1(d) and Table 2.1(d)
(863 msec and 897 msec, respectively, p <0.01) while there was no visual field
difference for Kana (883 msec for LVF and 886 msec for RVF, p >0.05).
Figure 2.2(d) indicates a great deal of individual differences in both Kanji and
Kana. That is, seven subjects showed a RH advantage and five subjects showed a LH
advantage for Kanji, while only three subjects showed a LH advantage, five showed no
difference, and four showed a RH advantage for Kana. There was one subject who
showed an extreme RH advantage for Kanji(about 160 msec). When this subject was
excluded from the analysis, the character-visual field interaction was not significant
(F(1.10) = 2.37, P > 0.05).
factors of visual field and character type showed that there was a main effect of
Error rate: The overall error rate was about I%. An ANOVA with two within
character type, 1.6% for Kanji and 0.5% for Kana, (F(I, ll) = 9.58, p < 0.05). The
ANOVA showed no visual field difference (F(I, ll) < I) and no interaction between
character type and visual field (F(I, ll) < I).
35
Insert Figure 2.2 (d)
The error rates indicate that the subjects made slightly more errors in Kanji
than in Kana. In fact, eight of 12 subjects made no errors in the Kana presentation
while only two of them were errorless for Kanji. It seems that the Kana task was
easier than Kanji task when the picture is clearly presented in both hemispheres.
However, since the error rates for Kanji and Kana were both very small, I do not
consider the difference between them revealing. Again, there is no speed-accuracy
trade-off, so the discussion is based on the reaction time results.
Discussion
The reaction time results indicate that Kanji words were responded to faster in
the RH than in the LH while Kana words did not show any hemispheric difference.
This result was different from the results from Exp. 3. When the picture was presented
for a short duration in the center of the visual field in Exp. 3, Kana characters showed
a LH advantage (11 among 12 subjects showed this tendency), while Kanji characters
did not show any overall hemispheric difference (four of 12 subjects showed LH
advantage, four showed no difference, and four showed RH advantage). However,
when the picture was presented for a longer duration in the center of the visual field
in Exp. 4, the result was opposite. The very large subject vanability described in the
Results suggests that the significant LVF advantage in Kanji in Exp. 4 might be due to
chance. In fact, without the one subject with an extreme RH advantage for Kanji, the
interaction was not significant.
It seems that when a picture is presented in the center of the visual field, Kanji
characters are processed efficiently in either hemisphere, depending on the individual,
while Kana characters are processed better in the LH when the previous picture
stimulus is presented for a short duration, but are processed equally wellin both
hemispheres when the picture is presented for a long duration. The latter result stands
against the hypothesis that Kana characters are only processed in the LH.
When a picture is clearly presented for a long time with long ISI, both
36
hemispheres might be ready to process both characters equally efficiently. Even Kana,
which usually requires phonological processing, may be processed in the RH by
matching the image of an expected character corresponding to a picture with the real
character presented since there is enough time for a subject to generate an expected
character internally. In this case, a visual matching strategy might have been used. In
fact, individual data showed that subjects who showed a RH advantage for Kana
tended to show a strong RH advantage for Kanji(Figure 2.2 (d)). Therefore, these
subjects might have used the same strategy to process Kanji and Kana characters. In
experiment 2, however, the peripheral presentation of the picture for 500 msec might
not have established quite such a clear representation of the picture. Therefore, the
visual matching strategy might have failed, and the subjects might have employed the
phonological strategy.
In summary, the results from the four experiments, in general, suggest that the
position and the presentation duration of the picture have a great impact on the
processing of Kanji and Kana characters' When the picture is presented at the side of
the visual field, differential processing of Kanji and Kana in the different
hemispheres is strongly manifested. When the picture is in the center of the visual
field, this differential processing seems to be weakened, and, in fact, individual
differences are found to predominate.
This pattern of results might reflect hemispheric activation (Kmsbourne, 1973).
As discussed earlier, the lateralized picture presentation seems to enhance the activity
of one hemisphere, and hence an incoming stimulus would be processed better in the
already activated hemisphere. Therefore, the interaction between character and visual
field would be increased. On the other hand, the centered picture presentation would
reduce or suppress the interaction because both hemispheres are already activated by
the picture. However, we do not know if the pictures used in these experiments are
actually processed preferentialIy in one hemisphere. In order to examine whether
picture processing itself does not bias one hemisphere over the other, picture
37
presentation durations needed to achieve the same performance level were measured in
each visual field in the next experiment.
The purpose of this experiment was to examine whether or not there is a
preferred visual field for picture recognition, and also to test whether the intensity of
stimulation makes a difference to any obtained visual field preference. I tested this by
varying the stimulus duration needed to make error rates equal in each visual field.
I also examined if a low stimulus intensity will yield an advantage in the LVF
Experiment 5: picture recognition
\
and high stimulus intensity will yield an advantage in the RVF. I asked this question
because it has been suggested that a stimulus with lower energy level produced by a
short exposure duration is responded to faster in the LVF, while a stimulus with higher
energy level produced by a long exposure duration is responded to faster in the RVF
(e. g. Pring, 1981). If the stimulus input level contributes to the visual field differences,
then the stimulus exposure duration for the LVF should be shorter in the low intensity
condition, whereas that for the RVF should be shorter in the high intensity condition,
for equal accuracy in the two visual fields.
undergraduate students who participated in the experiment in order to obtain an extra
Subjects: Subjects were 9 female and 7 male Northeastern University
credit for an Introductory Psychology course. All of the subjects were right handed on
the basis of the modified Edinburgh Handedness Inventory (see Bryden, 1977; Oldfield,
1971). None of their immediate family members were left handed. All but one subject
had at least 20125 corrected vision for each eye. The one subject had 20130 vision for
both eyes. The subjects were randomly divided into two groups: the low stimulus
38
Method
intensity group and the high stimulus intensity group.
Apparatus: The apparatus was the same as in the previous experiments.
pictures were the same stimuli used in the previous experiments. (See the description of
stimulus section in Experiment I. ) Stimuli were 1.5 degrees of visual angle in width
Stimulus: Stimuli were 40 line-drawing pictures of common objects. These
and height. The stimuli were rear projected onto the screen set 66 cm away from the
subject at eye level. Each stimulus appeared at 5 degrees of visual angle to the left or
right of the central fixation point depending on the condition. The luminance of the
background was 27 ft. -L. for the low intensity group and 41 ft. -I. for the high intensity
group.
Procedure: Each subject was instructed to sit at a table facing a translucent
screen with his/her head on a chin-rest. For each trial, the subject was asked to fixate
on a small fixation cross at the center of the screen for 2 seconds.
For each trial, a fixation cross was presented for 2 seconds, followed by a blank
screen for 500 milliseconds, and then one of the 40 pictures was presented. The subject
viewed the stimuli monocularly in either the LVF or the RVF. Monocular vision was
used as in lones and Aoki(1988), because the stimuli were projected onto the nasal
half field of the retina for both eyes, which is thought to have better resolution (See
Hirata and Osaka, 1969). To counterbalance the order of presentation, half of the
subjects viewed the stimuli in the LVF first with his/her left eye, and the other half
viewed the stimuli in the RVF first with his/her right eye. The subject had 20 practice
trials to become familiar with the task prior to each visual field test. Each test
consisted of 80 trials, in which the same picture appeared twice with either a cross or a
numeral as a fixation point.
The subject's task was to name a picture as fast and accurately as possible. The
initial stimulus duration was 200 milliseconds. This was increased by 10 milliseconds
39
following an incorrect response or decreased by 10 milliseconds following two
consecutive correct responses. The stimulus duration, the incorrect response, and the
reported fixation number were recorded.
There was a five minute break between the two conditions. The experiment took
approximately one hour,
Approximately 2% of the data for each subject was discarded because of incorrect
numeral responses. Each dependent variable was analyzed by 3 way Analysis of
Variance (intensity, order of presentation and visual field). The intensity and order of
presentation were between-subject factors, while the visual field was a within-subject
The data used for the analysis were stimulus durations and error rates.
factor.
Results
duration analysis (because of unstable responses). Consequently, the means of stimulus
durations for 70 trials in each condition were used for the analysis. The ANOVA
showed no main effects. That is, there was no overallsignificant difference between
the low and high intensity groups, 1/7 msec and 1/2 msec respectively (F(I, 12) < I), no
overall significant difference between orders of presentation (F(I, 12) < I), and no
significant difference between the LVF and RVF, 1/7 msec and 1/1 msec respectively
(F(I, 12) < I). There was no significant interaction of visual field with intensity
(F(I, 12) < I). The analysis, however, showed a significant interaction of visual field
with order of presentation (F(I, 12)=12.98, p<0.01). The three way interaction of visual
field, order of presentation and intensity was not significant (F(I, 12)=1.42, p > 0.05)
(see Figure 2.4).
A post hoc comparison indicated that there was a practice effect (p<0.05). That
is, the subjects who started with the LVF showed a shorter stimulus duration for the
Stimulus duration: The first 10 trials were excluded from the overallstimulus
40
LVF than the RVF, and vice versa for the subjects who started with the RVF.
However, the overall effect of order of presentation did not reach significance.
Therefore, counterbalancing the order of presentation eliminated the practice effect.
alone, showed the same tendency. (See Figures 2.6 and 2.7. )
The mean stimulus durations for the last 40 trials alone, and the last 20 trials
Error rates: Accuracy was equated at around 81% for all conditions by design,
and this was confirmed by ANOVA. The ANOVA showed no visual field difference,
18.1% for the LVF and 17.7% for the RVF (F(I, 12) < I), and no intensity difference,
17.7% for the low intensity group and 18.1% for the high intensity group (F(I, 12) <).
The interaction between the visual field and the intensity was not significant, either
Insert Figure 2.4
Insert Figure 2.6 and 2.7
(F(I, 12) < I). In the low intensity group, the error rates for the LVF and the RVF
were 18.6% and 16.8%, respectively. In the high intensity group, the error rates for the
LVF and the RVF were 17.5% and 18.6%, respectively. (See Figure 2.5). This non-
significant difference in error rates demonstrated the success of the procedure
employed here. That is, we tried to measure the difference in the input level of the
stimuli(stimulus durations) by keeping the performance level(error rates) equal.
41
achieve a certain level of performance were the same for both visual fields. The
The results from this experiment suggest that the stimulus durations required to
stimulus duration needed to achieve approximately 80% correct responses was about 1/5
Insert Figure 2.5
Discussion
msec (117 msec for the LVF and 112 msec for the RVF). Moreover, the stimulus
intensity had no effect. Therefore, the pictures used here were not lateralized overall.
In addition, these results imply that the pictures were at least 80% correctly
recognized in the previous experiments I and 3 with a short presentation (150 msec).
There is the possibility that a RH advantage for recognizing a picture might
have been canceled out by a LH involvement required by naming in this Experiment.
However, since the same pictures and same recognition strategy was required in the
earlier Experiments, the results indicate that picture recognition per se is not the cause
of the laterality effect I found in Experiments I and 2, but rather that character
recognition is lateralized.
A series of experiments in this chapter supported the hypothesis that Kanji and
Kana characters are processed differently in different hemispheres. More specifically,
in the semantic task, Kana characters are clearly processed faster in the LH than in the
RH while Kanji characters are processed faster in the RH than the left. This
dissociation is clear when both picture and word were presented to the same side of the
visual field, as in Exp. I and Exp. 2. When the picture was presented in the center of
42
General Discussion
the visual field and the word to the side, however, the hemispheric differences
weakened. When the picture duration was short, the hemispheric differences in Kanji
disappeared whereas those in Kana still remained. On the other hand, when the
picture duration was long, the hemispheric differences in Kana disappeared whereas
those in Kanji became marginalIy significant. Therefore, it is suggested that the
location and duration of the picture play an important role in manifesting the
hemispheric differences in processing Kanji and Kana.
Figure 2.2 shows the correlation across individuals of visual field effects for
Kanji and Kana. The abscissa represents the difference of reaction times between
RVF and LVF for Kanji, whereas the ordinate shows the same difference for Kana.
Positive numbers indicate a RH advantage (the LVF is faster than the RVF) while
negative numbers indicate a LH advantage (the RVF is faster than the LVF). For
Kanji, the majority of the subjects showed RH advantages in Exp. I and Exp. 2, while
some subjects showed RH advantages and the other showed LH advantages in Exp. 3
and Exp. 4. Since as almost equal number of the subjects in Exp. 3 showed right as left
hemispheric advantages, no overall hemispheric advantage was seen. Similarly, some
subjects showed a LH advantage in Exp. 4. However, since the magnitude of the
advantage for the RH was larger than that of the LH, and there was one strongly RH
oriented person, the overall result indicates a RH advantage. After excluding this
person, the character-visual field interaction was not significant. For Kana, the
majority of the subjects showed LH advantages in Exp. I, Exp. 2. , and Exp. 3, while most
subjects showed no hemispheric advantage in Exp. 4.
In summary, when a picture was presented to the side of the visual field, most
subjects showed a RH advantage for Kanji and a LH advantage for Kana. (See Figure
2.8(a)) When a picture was presented in the center of the visual field, some subjects
showed a RH advantage and the others showed a LH advantage for Kanji. (See Figure
2.8(b)). A majority of the subjects showed a LH advantage for Kana when the
duration of picture presentation was shorter, whereas most subjects showed no
43
hemispheric advantage when the duration of the picture was longer. The different
results for the short and long durations of picture presentation will be discussed in a
later section. The comparison between the two picture locations is depicted in Figure
2.8(a) and 2.8(b).
center). Although durations of picture presentation were different the data were
I examined the magnitude of interaction between the two presentations (side vs.
Insert Figure 2.8(a) and 2.8(b)
combined for lateralized picture presentation (Experiment I and 2) and for centered
picture presentation (Experiment 3 and 4), since there was no significant difference
between Experiments I and 2, nor between Experiments 3 and 4 (separate ANOVAs
showed F(1.22)<I for both pairs). The difference of reaction times between RVF and
LVF (visual field advantage) was used as a dependent measure. A 2-way ANOVA with
picture presentation (lateralized vs. centered) and characters type (Kanji vs. Kana)
showed no difference in picture presentation condition (F(1.46)<I), but showed a
significant main effect of character type (F(1.46)=47.6 P<. 001) and a significant
interaction between presentation condition and character type (F(1.46)=4.55 P<. 05).
Kanji and Kana showed opposite direction of advantage (20 msec of a LVF advantage
for Kanji and 33 msec of a RVF advantage for Kana). Post hoc tests showed that the
advantages for both Kanji and Kana were larger in the lateralized picture condition
than in the centered picture condition (28 msec and 13 msec for Kanji, and 41 msec
and 24 msec for Kana, respectively). Therefore, the character-visual field interaction
was larger in the lateralized picture condition than in the centered condition.
There is a possible interpretation which may explain why the location of the
picture plays a role in the character-visual field interaction. As described in Chapter
I, Kirisbourne's activation theory predicts that activation of the LH should enhance
the processing of stimuli which require phonological(LH) processing, such as Kana.
However, the activation of the RH, in this case, does not help efficiency of processing
44
the stimulus because the stimulus needs to be transferred to the LH. Consequently, this
case yields a clear difference in performance between the two hemispheres as seen in
Exp I and Exp 2 for Kana. In the same way, if the stimulus is processed better in the
RH (due to visual spatial processing), then activation of the RH should yield a greater
advantage than activation of the LH. This is the case for Kanji as seen in Exp I and
Exp 2. Therefore, a lateralized picture should enhance the character-visual field
interaction.
How about the presentation of the picture in the center? In this case, the
picture activates both hemispheres. When both hemispheres are activated, they are
ready to process any incoming stimuli efficiently and the laterality effect would be
reduced or suppressed. The character-visual field interaction should be reduced as in
Experiments 3 and 4. This is clearly seen in Experiment 4. However, in Experiment 3,
Kana still showed a LH advantage (11 out of 12 subjects showed this trend). This
might be due to the short duration of picture presentation. When the picture duration
is longer, there is enough time for both hemispheres to process the stimuli equally.
Indeed, the subject might have used visual imagery generated by the picture to permit
a visual matching strategy as well as phonological strategy for Kana. In Experiment 3,
when the picture was presented for a short period, the subject might not have enough
time to generate a useful image, and so phonological processing dominated.
The overall results from the experiments above revealed that Kanji characters
were processed better in the RH than in the LH while Kana characters were processed
better in the LH than in the RH, in the semantic task, when one hemisphere was
activated by the previous picture presentation. This is inconsistent with the findings
from the previous studies by Hayashi and Hatta (1983) and Hatta (1981). They found
that Kanji characters were processed better in the LH in the case of a semantic task.
Hatta (1981, 1983) suggested that the RH processed Kanji better in the case of physical
identification whereas the LH processes Kanji in the case of semantic decision. Hatta
(1981a) also found no significant difference between the LH and the RH in the case of
lexical decision. Based on these results, Hatta concluded that higher levels of linguistic
45
processing such as semantic decision require LH function. However, the semantic tasks
used so far were a semantic categorization task and a semantic congruency task, which
require more abstract processing and a step beyond the comprehension of word
meaning. The semantic task used in this study is a concrete semantic task which
requires matching the meaning of a picture with a word. The pattern of results
showed that the LH can be better for this task in Kana whereas the RH can be better
in Kanji, at least when a picture and a word are both presented successively in the
same hemisphere.
A semantic comparison task, in which subjects judged whether the physicalsizes
of two concrete Kanji characters were congruent with the relative sizes of reallife
objects, yielded a LVF-RH advantage (Hatta, 1983). Hatta suggested that in this task
subjects used imagery in the semantic judgment. The use of imagery may also account
for some of the results in Experiment 4 as discussed earlier, but we have no direct
evidence on this point.
study only used highly concrete, highly familiar, and highly hieroglyphic characters,
whereas Hatta (1981a) and Hayashi and Hatta (1982) used abstract characters as well as
concrete characters, Therefore, the stimuli used in this study might be more favorable
Another difference between the previous studies and this study is that this
to the RH. If we used more abstract characters, the results might be different.
However, concreteness alone cannot explain the results because the Kana words
transcribed from the Kanji words showed the opposite results. Therefore, some other
aspect of Kanji, such as pictorialness and familiarity or the level of mapping from
print to sound, must favor the RH.
The Kana words used in this study were unfamiliar to Japanese readers since
the Kana words were transcription of the Kanji words. Reading unfamiliar words
must be mediated by phonological processing (e. g. Bresner and Hilderbrandt, 1987;
Waters and Seidenberg, 1985). If this is the case, unfamiliarity of Kana words alone
might have favored the LH. In order to test this hypothesis, one could use Katakana
characters (square scripts) which are normally used for loan words, such as television
or radio. Besrier and Hilderbrandt (1987) suggested that lexical access of Katakana
familiar words can be achieved without reference to phonology. If familiarity alone in
46
this study favor the RH, then Katakana familiar words in a picture-word matching
task would yield a RH advantage. However, if some other factors than familiarity is
responsible for the results in this study, then Katakana words also should yield a LH
advantage.
In the next chapter, only Kanji characters were used to control for visual
familiarity. Kanji characters were tested in both semantic and phonological tasks. I
expected an interaction between task and visual field.
47
Chapter 3: Semantic and Phonological Processing of Kanji
In this chapter, I examine if the right hemisphere has the capacity to process
Kanji characters semantically. Kanji characters were used to examine whether the
same characters are processed differently in the two hemispheres depending on the
processing demands.
It has been suggested that the same stimulus may be processed mainly in the RH
when a visual or perceptual decision is required and in the LH when a verbal or
categorical decision is required (See Goodglass and Butters, 1988). For example, Bryden
and Allard (1976) demonstrated that alphabet letter naming was better in the LH when
the type face was simple block letters, and it was better in the RH when the type face
was perceptually difficult. Presumably the perceptual difficulty of the complex type
face involves more visual processing, which is so much better in the RH that a RH
advantage ensued, while simple block letters require less visual processing so the basic
LH dominance for verbal tasks asserted itself. Cohen (1972) found that physical
matching of uppercase letters (e. g. AA) was better in the RH, while matching an
uppercase with a lower-case letter (e. g. Aa) was better in the LH. Sasanuma at al.
(1980) also found that a visual matching task tended to yield a RH advantage and a
phonological matching task tended to yield a LH advantage for both Kanji and Kana.
Therefore, a task demand or processing strategy can shift the hemispheric advantage
Introduction
48
I
obtained for the same stimulus.
If the RH has the capacity to comprehend Kanji words, a picture-word
matching task might yield a RH advantage when the match is by meaning. However,
the same Kanji words might be processed better in the LH when implicit phonological
processing is required, as in the case of matching by sound. The picture-word
matching task is useful for comparing performance in the semantic task with
*
performance in the phonological task because one can use the same set of stimuli for
both tasks, without requiring overt phonological processing. In the tasks used here, the
set of stimuli consisted of four types of Kanji characters: a) standard, b) homophone, c)
semantically related, and co unrelated. The standard character matches a picture in
both meaning and sound. The homophone matches a picture in sound only but not in
meaning. The semantically related character matches a picture in meaning, if
mexactly, but not in sound. The unrelated character does not match a picture either in
sound or in meaning. In the semantic task, one would be asked to match the meaning
of a picture with the meaning of a word, so one would choose the standard and
semanticalIy related characters as correct responses. In the phonological task, one
would be asked to match a name (sound) of the picture with a sound of the word. In
this case, one would choose the standard character and homophone as correct responses.
If semantic processing takes place in the RH, the semantic task should yield a RH
advantage when subjects pay attention to meaning and ignore sound. In the
phonological task, however, the balance should shift towards the LH since only the LH
has phonological function (e. g. Zaidel & Peters, 1981).
Three experiments using the picture-Kanji matching task were conducted in this
study. In Experiment I, a picture was presented in the center of a visual field in order
for the picture to be fully represented in both hemispheres. A Kanji character was
then presented to either the left or RVF. In Experiment 2, both pictures and Kanji
characters appeared in the same side of the visual field, using an ABBA design.
Experiment 3 replicated Experiment 2, but used an AABB design, in which the two
tasks (2 sessions for each) were conducted on two different days (a week apart).
49
phonological task just described. If both semantic and phonological processing for
Kanji take place solely in the LH, then reaction times in the RVF should be faster
than those in the LVF regardless of the task. However, if the RH has an advantage in
processing Kanjisemantically, then reaction times in the LVF should be faster than
those in the RVF in the semantic task, and vise versa in the phonological task.
Experiment 6: Picture-Kanji Matching Task
In this experiment, subjects performed the two tasks, a semantic task and a
with a centered picture
Boston area, and often spoke and read Japanese. The average age of the subjects was
Subjects: were 32 Japanese people (16 males and 16 females) who resided in the
about 26 (28 for male and 24 for female), ranging from 18 to 33 years of age. One
half of the subjects had stayed in the U. S. for less than I year while the other half had
stayed in the U. S. for more than 2 years, These two groups of subjects were examined
separately to consider any second language effect due to the use of English, since
50
Tsunoda (1985) reported that the use of a second language affects the shift of
Method
hemispheric superiority in auditory experiments.
All the subjects were right-handed according to the standard simplified
Edinburgh Handedness Inventory (See Bryden, 1977; Oldfield, 1971), and none of their
immediate family members were reported left handed. Anthe subjects had normal or
corrected to normal vision (at least 20120) binocularly as assessed with the use of a
Snellen chart prior to the experiment. The subjects were paid $6.00 per hour to
participate in the experiment.
Apparatus: A 19" TV monitor (Ikegami Model TM20-9RHA) driven by a DEC
PDP 1/173 computer with ITl frame buffers and routines was used in place of a
tachistoscope. The monitor refreshed the screen at a rate of 67 Hz, and was set I in
away from subject's eye. A video camera (Panasonic Model No. AGE-100) was used to
take pictures of stimuli which were digitized and stored in the computer's memory. A
set of six response keys were connected to the computer in three pairs, with the keys in
each pair wired in series. The response keys were mounted on a wooden box, three on
the left and three on the right. Subjects started a trial by simultaneously hitting the
two lowest keys with the right and left thumbs. At the end of each trial, they
indicated "yes" or "no" by hitting the middle or top buttons with their index and
middle fingers. Simultaneous bimanual responses were required in this experiment in
order to eliminate biased activation of motor areas in one of the hemispheres. The
results, therefore, could be interpreted on the basis of either sensory/input processing
or cognitive processing but not motor/output processing. Half the subjects used the
index finger for "yes" response and the middle finger for "no" response, while the other
half used the opposite fingers. Reaction times were measured from the onset of the
Kanji character to the subject's key press.
51
Stimuli: Stimuli were 16 line drawings of familiar objects and 64 single Kanji
characters, The Kanji characters were divided into 4 categories: 16 standard characters
(matching the pictures in both sound and meaning), 16 homophones (matching the
pictures in sound but not in meaning), 16 semantically related characters (matching the
pictures in meaning but not in sound), and 16 unrelated characters (not matching the
pictures in either sound or meaning). All characters, except for some homophones,
were highly concrete and highly familiar, based on Kitao at al. 's (1977) list of 881 basic
Kanji characters, About one third of the homophones were abstract or less familiar in
order to match the sounds of particular pictures. (However, this should not influence
the results since the same stimuli were used in each task. )
All stimulus were 1.5 degree of visual angle in both height and width. The
pictures were presented in the center of the visual field while the Kanji characters
were presented at 5 degree of visual angle, either left or right of the center. The
luminance of the background screen was approximately 32 ft. -L. The contrast of the
stimuli was low, and varied over subjects (see Procedure).
The Kanji characters were photocopied from a Japanese dictionary (Kadokawa
Shin Kokugo Jiten, 1981) and were enlarged to approximately 2 cm x 2 cm. The
pictures were hand-written line drawings within 2 cm squares. The Kanji characters
and the pictures were arranged on a white sheet, videotaped, thresholded to make the
background uniformly light and the stimulus uniformly dark, and stored in the
memory of the computer.
Tasks: I) Phonological task: The subjects decided if a picture and a word were
matched or not on the basis of their pronunciation. They were told to ignore the
meaning as far as possible. The sixteen standard characters and the 16 homophones
were considered as matched, while the 16 characters related semantically to the
pictures and the 16 characters unrelated to the pictures were considered as non-
matched (see Figure 3.0).
52
2) Semantic task: The subjects decided if a picture and a word were matched
or not on the basis of their meaning. They were told to ignore the pronunciation as
far as possible. The words used in this task were the same as the characters used in
the phonological task. In this task, however, the 16 standard characters and the 16
semantically related characters were considered as matched, while the 16 homophones
and the 16 unrelated characters were considered as non-matched (see Figure 3.0).
Insert Figure 3.0
supported to minimize head movement. Prior to the experiments, the subject saw a set
Procedure: A subject sat in front of the screen with his/her chin and forehead
of picture stimuli on the screen, and was instructed to name the pictures and to
remember their shape and name. This was necessary to make sure that the subject
identified and named the pictures correctly. If the subject named a picture incorrectly,
he/she was corrected.
The subject then performed two sets of practice sessions, each consisting of 28
trials. One set involved a semantic task and the other a phonological task. The subject
was fully informed about the two tasks, in Japanese, to reduce the chance of confusion.
practice trials in order to make the overall error rate approximately 10%. (This error
rate was considered sufficiently high to permit detection of any speed-accuracy trade-
off (SATO).) The final contrast of the Kanjistimuli was then determined for each
The contrast of the Kanjistimuli was varied from 3% to 30% during the
individual, and was between 5% and 10%. After the practice trials, four sessions of 64
trials were performed. The four sessions consisted of two sessions with the semantic
task and two sessions with the phonological task. Each character appeared once in each
visual field at random, resulting in 128 trials in each task. Each task was divided into
two sessions of 64 trials. The 64 picture-character pairs were randomly selected by the
computer for each subject.
On the beginning of each trial, the subject was presented with a cross in the
center of the screen. After the subject fixated on the cross, he or she began the trial
by pressing the lower pair of response key. A picture was presented in the center of
the visual field for I second, and then, following a I second interstimulus interval(ISI)
with a fixation cross, a Kanji character appeared in either the left or RVF for 150
msec. The subjects then pressed two of the four keys to indicate whether the picture
and the Kanji character matched or not.
53
Reaction times and error rates were collected and recorded by the computer.
The experiment took approximately I hour,
Design: This was an ABBA design, in which half the subjects started with the
first half of the semantic task, followed by the two sessions of the phonological task,
and ended with the second half of the semantic task. The other half started with the
first half of the phonological task, followed by the two sessions of the semantic tasks,
and ended with the second half of the phonological task. The reason for using this
design was to establish coinparable reaction times for the two tasks, since a pilot study
showed a practice effect throughout alithe trials. There were breaks of a few minutes
between the four sessions.
Reaction times: A preliminary analysis showed no sex difference (F(I, 28) < I).
Therefore, data were collapsed over sex in the following analyses. Reaction times were
logarithmically transformed prior to the analyses. This made the positively skewed
distributions found in each data set approximately normal, and also made the variances
roughly homogeneous across data sets (see Kirk, 1982). The means were transformed
back to reaction times in milliseconds by taking antilogarithms, so that the means
indicated in tables and graphs are geometric.
A 4-way ANOVA was carried out with one between-subject and three within-
subject factors; the between-subject factor was the length of stay in the U. S. (less than
one year vs. more than two years). The within-subject factors were task (phonological
vs. semantic), visual field (left vs. right), and character type (standard, homophone,
semantically related, and unrelated).
An ANOVA performed on correct responses showed significant main effects of
54
Results
task (F(I, 30)=18.98, p<. 001), visual field (F(I, 30)=4.23, p<. 05), and character type
(F(I, 30)=110.05, p<. 001). However, the length of stay in the U. S. was not significant
(F(I, 30)<I, p>. 05). The ANOVA also showed a significant interaction between task and
character type (F(3,90)=32.9, p<. 001). A marginal four-way interaction between length
of stay, task, visual field and character type (F(3,90)=2.48, p=. 07) was difficult to
interpret and is not discussed further. None of the other interactions were significant.
The following describes the four significant effects in detail. For the task
difference, the mean reaction time in the phonological task was faster than that in the
semantic task (820 msec and 867 msec, respectively). (See Figure 3.1(a).) This is
different from what might be expected. If the access to the Kanjilexicon is not
necessarily mediated by phonological processing and is achieved by direct visual
processing, then the semantic task should be faster than the phonological task. Instead,
the result suggests that phonological activation might have occurred more rapidly than
semantic activation. This is supported by the analysis, restricted to the standard
characters alone, which still showed a significant task difference (F(I, 30) = 20.29, p <
0,001), with the phonological task faster than the semantic task (697 msec and 746
msec, respectively).
This overalltask difference might be due to task difficulty. That is, decision-
time for semantic relatedness might have taken longer than that for hornophonicity.
However, this may not be the case because reaction times for the semantic related
characters were not longer than other characters in the semantic task. (See Figure
3.2(b).) However, there is the possibility of a SATO discussed in the error rate section
55
below.
LVF (LVF) was faster than that in the RVF (RVF) (836 msec and 850 msec,
As for the significant visual field difference, the mean reaction time in the
respectively, p<0.05). (See Figure 3.1(b).) Both the semantic and phonological task
showed this tendency as seen in Figure 3.1(c). The mean reaction times for the RVF
and LVF in the semantic task were 871 msec and 861 msec, respectively, and those in
the phonological task were 830 msec and 811 msec, respectively. Although the
differences were small, the data suggest a LVF advantage overall. This result supports
the results of experiments in Chapter 2, which showed a RH advantage for processing
Kanji. This suggests that the RH is more involved in processing Kanji when overt
phonological processing is minimized.
responded to faster than were the homophones, the semantically related characters and
the unrelated characters (703 msec, 881 msec, 906 msec and 899 msec, respectively,
As for the differences in the character type, the standard characters were
p<0.01 with the Newman Keuls test). (See Figure 3.2(a)). There was no significant
difference between the latter three character types. This tendency was observed in
both visual fields (see figure 3.2(c)). This might suggest that the response times were
faster when information about both meaning and sound of a Kanji character matched
those of a picture than when either of these did not match, perhaps because the picture
activated both the meaning and sound, and facilitated character recognition. In other
words, the subject might have expected the standard characters due to picture priming.
As for the interaction between task and character type, nearly all pairwise
comparisons were significant at 0.05 level by the Newman-Keels test (see Figure
3.2(b)).' This was interpreted as follows.
First, the mean reaction time in the phonological task was generally faster than
that in the semantic task, except for the semantically related characters, where the
reverse occurred (p< 0.05).
56
interference took place in the phonological task when semantically related words were
presented.
phonological task) usually take longer than "yes" responses (the semantic task).
However, a more detailed comparison supports the same conclusion. In the
This comparison is not fair by itself, because "no" responses (the
phonological task, the mean reaction time of the semantically related words was
longest, and was significantly slower than that of the unrelated words. Since both the
I. The following four pairs were not significant: the semantically related charactersin the semantic task and the unrelated characters in the phonological task, theunrelated characters in the semantic task and the semantically related characters inthe phonological task, the unrelated characters in the semantic task and thehomophones in the semantic task, and the semantically related characters in thesemantic task and the homophones in the semantic task.
This interaction seems to indicate that semantic
semantically related words and the unrelated words were "no" responses in this task, the
only difference between two types of character which might have caused a significant
delay for the semantically related words is the semantic relatedness to the picture.
Therefore, one may claim that semantic similarity interferes with the phonological
decision. Thus, even when the phonological aspect of the character is emphasized in
the task, semantic information interferes with processing, so that the reaction time
increases for the semantically related words.
Second, there is no phonological interference based on the results from the
semantic task. In the semantic task, the reaction times to the homophones and the
unrelated words did not differ, but they were longer than the reaction times to the
semantically related words and the standard words. Since the homophones and the
unrelated words are both "no" responses, and the only difference between the two is
that the former has the same sound as the picture while the latter does not, it is
reasonable to say that there is no phonological interference or facilitation. Similarity
in sound neither helps nor hinders the judgement of the meaning of Kanji characters,
(Note: although the semantically related words were responded to faster than were the
homophones and the unrelated words, the former were "yes" responses and the latter
two were "no" responses, so we cannot determine from these data if there is any
57
semantic facilitation. )
Third, when the two "yes" responses were compared, the standard characters
were responded to faster than the others in both the semantic and phonological tasks.
This effect may be explained by a cognitive activation hypothesis. Even though the
pictures activate semantically related words, the degree of activation may be modulated
by semantic relatedness, which can vary across the characters, Also, when the pictures
activate homophones, the number of homophone is typically quite large in Japanese, so
that amount of activation for each character may be relatively small. Therefore, it
takes some time to judge the semantic relatedness or sound similarity of these
characters, resulting in longer reaction times. On the other hand, a standard character
may generate relatively more activation because a picture primes both the meaning and
sound. Consequently, it should take less time when the character is activated by both
types of information than when activated by only one type of information.
Error rates: The overall error rate was approximately 10%. A preliminary
analysis showed no significant sex effect. Consequently, sex was collapsed for the
following analyses. A four-way ANOVA with one between-subject and three within-
subject factors was performed on the error rates. The between-subject factor was
length of stay in the U. S. , and the within-subject factors were task, visual field and
character. The same analysis was performed on arcsine error rates. Since the two
analyses produced similar results, the results from the untransformed error rates are
Insert Figure 3.1(a),(b),(c)
Insert Figure 3.2 (a),(b),(c),(d)
discussed below.
58
Two main effects were found to be significant: the task (F(I, 28) = 18.89,
p<0,001) and the character type (F(3,84)=26.44, p<0,001). First, the subjects made fewer
errors in the semantic task than in the phonological task (8% and 12%, respectively).
(See Figure 3.3(a).) This indicates that the faster response times found in the
phonological task may be due to a speed-accuracy trade-off (SATO), although a full
SATO curve would be needed to firmly establish this point. Second, the subjects made
fewer errors with the standard and unrelated characters than with the homophones and
semantically related characters (6%, 6%, ISVo and 13%, respectively). (See Figure 3.4(a).)
This tendency was seen in both visual fields. (See Figure 3.4(c).) This result suggests
that the homophones and the semantically related characters were effective distractors.
There was no significant difference between the standard characters and the unrelated
characters, nor between the unrelated characters and the homophones.
There was no significant main effect of visual field (F(I, 28)<I), or of the length
of stay in America (F(I, 28) < I). There was also no significant interaction between
task and visual field (F(I, 28) < I). The error rates in the two visual fields were not
different for either task. (See Figure 3.3(c).)
(F(3,84)=21.82, p<0,001). The Newman-Keuls comparison test indicated the following
significant results. In the phonological task, the subjects made more errors with the
homophones (23%) than with the other three character types, as seen in Figure 3.4(b)
(p<0.01). In the same task, the semantically related characters (16%) yielded more
errors than the standard (6%) and unrelated characters (3%) (p<0.05). However, in the
There was, however, a significant interaction between task and character
semantic task, there was no significant difference between the character types. This
tendency was seen in both visual fields. (See Figure 3.4(d).) These result indicate that
the subjects might not have considered some of the homophones as homophones. A
follow-up analysis showed the high error rate with the homophones might have been
due to the difficulty of a few particular items. After these were omitted from the
analysis, the character-task interaction was not significant (p<0.05), although the
pattern of both error rates and reaction time were stinthe same as the pattern prior to
59
omission of the items.
One interesting feature is that in the phonological task the semantically related
characters yielded more errors than did the unrelated characters (p<0.05). This result
confirms the semantic interference in the phonological task shown in the reaction time
data. Thus, even when the subjects paid attention to the sound of the Kanji characters,
semantic relatedness interfered with their judgements. In contrast, there was no
phonological interference in the semantic task since no significant difference was
found between the homophones and the unrelated characters in the semantic task. This
is also consistent with the reaction time data.
The unrelated characters yielded more errors in the semantic task than in the
phonological task (p<0.05). This may be due to the fact that some of the unrelated
characters, such as "dog" and "inn", were considered semantically related to the pictures
of teeth and fish for the Japanese subjects. When these characters were replaced in
Experiment 2, the difference between the homophones and unrelated characters
disappeared.
There were no other significant interactions.
task and visual field. However, there was an overall LVF advantage. Both the
The results from Experiment 6 showed that there was no interaction between
semantic and phonological tasks tended to show slight LVF advantages. This may
mean that Kanji characters are processed better in the RH regardless of tasks. This
result is inconsistent with some of the previous studies. Sasanuma at a1. (1980) showed
that a phonological task (sound matching) yielded a RVF advantage for Kanji
characters while Hatta (1981a) and Hayashi and Hatta (1982) showed that a semantic
task (semantic congruency or semantic categorization) also produced a RVF advantage.
This inconsistency might reflect any of four factors.
The first possible factor is procedural. Sasanuma at a1. (1980) used simultaneous
presentation of two Kanji characters in her phonological task, while the current study
employed successive presentation of a picture and a Kanji character. The use of the
picture might have biased processing somewhat towards the RH and so canceled the LH
advantage for phonological processing. Hatta (1981a) and Hayashi and Hatta (1982)
presented only one Kanji character in their semantic task, and asked a subject to judge
whether the meaning of the character was congruent with the location of the character
Discussion
60
or whether the character belonged to a particular category. On the other hand, the
current study used both picture and Kanji, and required a very low level of semantic
processing, i. e. comprehension of the meaning of the character. It is possible that
comprehension of Kanji characters is available in both hemispheres, possibly slightly
better in the RH, while higher cognitive processing, such as semantic categorization,
takes place only in the LH.
The second possible factor is the use of highly familiar and highly concrete
Kanji characters in the current study. It is possible that familiarity and concreteness
might have facilitated the RH function (e. g. Eijis and Shepherd, 1974; Hines, 1977).
The other studies did not control for these factors.
The third possible factor is the use of a picture as a target. The picture might
function as a priming cue, which activates both hemispheres because the picture was
presented in the center of the field for long enough (I sec) with a long ISI (I sec).
Therefore, by the time the Kanji character was presented, both hemispheres might have
been activated equally and were ready to process any type of character. Consequently,
a hemisphere difference might not be observed.
Experiment 4 is similar to this experiment, in that a picture was presented in the
This idea is supported by the results from the experiments in Chapter 2.
center for I sec and a word was presented to the side for 150 msec. In Experiment 4,
the RVF advantage for Kana disappeared despite the fact that the other experimental
conditions showed a strong RVF advantage. One might speculate that the presentation
61
of the picture in the center for a full second might balance out the activities of the
two hemispheres for Kana, while stillshowing a LH advantage for Kanji.
hemispheric differences for processing Japanese characters exist when potential
Indeed, it was one of the purposes of this chapter to establish whether
hemispheric biases due to a lateralized picture are minimized. If the picture is
presented to either side of the visual field and activated only one hemisphere, then the
result is different. In fact, the series of experiments in Chapter 2, using both Kanji
and Kana, indicate that the location of picture seems to be crucial in yielding a visual
field effect.
the order of the two tasks in the ABBA design.
The fourth possible factor is the mode of presentation. This experiment mixed
influenced later ones, for example, the subjects might have acquired a strategy for the
first task and employed the same strategy to perform the other task. As a result, a
possible hemispheric difference might have been reduced.
performed. It was also interesting to examine whether semantic interference would be
observed in the phonological task but no phonological interference in the semantic
In order to test the third and fourth possibilities, Experiment 7 and 8 were
task, since this result was consistent for both reaction times and error rates in
Experiment 6. The interference results support the idea that access to the lexicon in
Kanji does not require phonological mediation but is achieved via direct visual codes
(i. e. , grapheme to morpheme). If so, access to the meaning took place earlier than
access to the sound in Kanji words, and the similarity of meaning might have confused
the judgement of sound, but the similarity of sound would not have interfered with the
judgement of meaning.
It is possible that earlier tasks
62
Kanji character is crucial in producing a hemispheric asymmetry. Based on the
tachistoscopic experiments in chapter 2, a visual field effect on reaction time was more
This experiment examines whether the location of a picture presented prior to a
Experiment 7: Picture-Kanji Matching Task
obvious when the picture was presented to the same side as the word than when the
picture was presented to the center.
with a lateralized picture
The hypothesis is that with the picture in the same visual field as the character,
the phonological task should yield a RVF advantage while the semantic task should
yield a LVF advantage; or, if there is an overall LVF advantage, the LVF advantage
for the semantic task should be greater than that for the phonological task.
Subjects: Subjects were 16 Japanese people (I male and 15 females) who had
resided in the Boston area for an average of three months (ranging from I month to 2.5
years). The length of stay in U. S. was not controlled in the next two experiments
because Experiment I did not show any significant effect of this factor. They often
spoke and read Japanese. The average age of the subjects was 20, ranging from 18 to
28 years of age. The number of males and females was not balanced since Experiment
I showed that there was no sex effect.
Method
All subjects were right-handed without any immediate left-handed family
member, and had normal vision based on the same tests used in Experiment I. The
subjects were paid $6.00 for a I hour session.
Apparatus: The apparatus was the TV monitor and computer used in
Experiment I. In addition, a video camera (Panasonic Model No. AG-100) was used to
make sure that the subjects fixated on the fixation point. Since the picture and the
Kanji character were presented to the same visual field, the subjects could easily
predict where the character would appear. Therefore, the eye monitor was necessary to
detect possible eye movement. The subject's eye movement was shown on the small TV
screen (RCA Model No. Tcl910) next to the experimenter while the experiment was
being conducted. The experimenter checked the trials in which eye movement was
observed and those trials were omitted from the data analysis. This happened only
63
0.4% of the trials.
few unrelated characters were changed since these characters were sometimes
Stimuli: Stimuli were the same as in Experiment I except for minor changes. A
considered as semantically related characters in Experiment I. One set of the picture-
word combinations was replaced because the homophone combination was too difficult.
Task: Tasks were the same as in Experiment I.
Procedure: The procedure was almost the same as in Experiment I. There were
two changes, however. First, the picture was presented to either the LVF or RVF for
240 msec, followed by ISI of I sec, and then the Kanji character was presented for 150
msec to the same side as the picture. The picture was presented for longer than the
character in order to make the picture clearly visible since subjects had earlier
reported that the pictures were more difficult to identify than the Kanji characters,
Second, reaction times were measured from the onset of the Kanji character to the time
when the subjects pressed and released the response keys with both hands.
Consequently, the reaction times of Experiment 2 were longer than those of Experiment
I, in which the time to the key press was measured.
Reaction time: Reaction times were logarithmically transformed prior to the
analysis for the same reason as in Experiment I. The means of these reaction times
were transformed back to reaction times in milliseconds to represent geometric means
in the graphs.
A 3-way 2 x 2 x 2 ANOVA was carried out on correct response times with 3
within-subject factors: task (semantic vs. phonological), visual field (right vs. left), and
character type (standard, homophone, semantically related, and unrelated).
The results showed significant main effects of character type (F(3,45)=40.08,
p<0.001), but there was no significant main effect of task (F(I, 15)<I) and visual field
(F(I, 15)<I). There was virtually no difference between the semantic and phonological
task (1690 msec and 1673 msec, respectively) and between the RVF and LVF (1681
64
Result
msec and 1682 msec). (See Figure 3.5(a) and 3.5(b).) There was 00 significant
interaction between task and visual field (F(I, 15)<I). The reaction times for the RVF
and LVF in both tasks were virtually the same as seen in Figure 3.5(c).
characters were significantly faster than to the homophones, the semantically related,
Newman-Keuls post comparison tests show that responses to the standard
and the unrelated characters (1459 msec, 1736 msec, 1772 msec and 1781 msec,
respectively). (See Figure 3.6(a).) There was no significant difference between the
latter three characters. This result is consistent with Experiment I.
(F(3,45)=4.50, p<0.01). A further analysis of the interaction using the Newman-Keuls
\
There was a significant interaction between task and character type
post comparison test showed that the standard characters were responded to faster than
the other characters in both tasks (p<0.05), and that the homophones were responded to
faster than the semantically related characters in the phonological task (p<0.05). The
overalltrend of Experiment 7 was similar to that of Experiment 6. That is, there was
a tendency for the semantically related characters to take longer than the unrelated
characters in the phonological task (18/7 msec and 1776 msec, respectively) although
this difference (41 msec) was not statisticalIy significant. There was no statistical
difference between the homophones and unrelated words in the semantic task (1802
65
msec and 1786 msec, respectively, a difference of 16 msec) (See Figure 3.6 (b).)
Therefore, the pattern shown in Experiment 6 (i. e. a semantic interference in the
phonological task and no phonological interference in the semantic task) also occurred
in Experiment 7, but was not statisticalIy significant. This lack of significance may be
due to a relatively small number of subjects, 16, as opposed to 32 subjects.
None of the other interactions were significant.
Insert Figure 3.5 (a),(b),(c)
analysis. However, since the results of ANOVA for the error rates and arcsine error
rates were almost identical, the results of ANOVA based on the untransformed error
Error rate: The error rates were converted to arcsine error rates for the
rates are discussed in this section.
A 3-way 2 x 2 x 4 ANOVA with 3 within-subject factors was employed for the
analysis. The 3 factors were task (semantic vs. phonological), visual field (right vs.
left), and character type (standard, homophone, semantically related, and unrelated).
The ANOVA showed significant main effects of task (F(I, 15)=20.4, p<0,001) and
Insert Figure 3.6 (a),(b),(c),(d)
character type (F(3,45)=19.29, p<0,001), but not of visual field (F(I, 15)<I). The
phonological task yielded significantly more errors than the semantic task (10.8% and
6.6%, respectively), which confirmed the results from Experiment I indicating that the
phonological task was more difficult than the semantic task (see Figure 3.7(a).) As for
character type, the Newman-Keuls test indicated that the semantically related
characters yielded more errors than the unrelated and standard characters (13.2 %, 3.8
% and 6.3 %, respectively) (p<0.05), but not than the homophones (11.5%), and that the
homophones produced more errors than the unrelated characters (p<0.05), but not more
than the standard characters (see Figure 3.8(a).) The test also showed that there were
significant differences neither between the homophones and semantically related
66
characters nor between the standard and unrelated characters, Therefore, in this
experiment, the homophones and semantically related characters worked as effective
distractors, as in Experiment 6. With respect to visual field, there was no difference
between the RVF and LVF (8.5% and 8.8%, respectively), (see Figure 3.7(b).)
The ANOVA showed a significant 2-way interaction between task and character
type (F(3,45)=10.82, p<0.001) (see Figure 3.8 (b)), but not between visual field and
character (F(3,45)<I) (see Figure 3.8(a)). A further analysis with the Newman-Keuls
test indicated that in the semantic task there was no significant difference between
any characters, On the contrary, in the phonological task the homophones and
semantically related characters produced more errors than the standard and unrelated
characters, However, there was no difference between the former two characters nor
between the latter two characters, As before, semantic information increased the errors
in the phonological task (compare unrelated with semantically related), but
phonological information had no effect on errors in the semantic task (compare
unrelated with homophones).
Based on the above results, it seems that similarity in both sound and meaning
does not influence the semantic judgement, while similarity in both sound and meaning
affects the sound judgement. Homophones yielded significantly more errors in the
phonological task than in the semantic task, which replicated the results in Experiment
6. This might be due to difficulty in deciding the sounds of some of the homophones
because they have multiple readings.
The analysis also showed a significant 3-way interaction between task, character
type and visual field (F(3,45)=3.35, p<0.05) (see Figure 3.8(d)). In both tasks, errors in
the most difficult conditions (homophones in the phonological task, and semantically
related characters in the semantic task) were slightly more frequent in the LVF(RH)
than in the RVF(LH).
67
First, there was no task-visual field interaction. Second, responses to the standard
The results from this experiment were similar to those from Experiment 6.
characters were faster than the other three character types. Third, there was a
tendency for semantic interference in the phonological task, but no phonological
interference in the semantic task. However, task difference and visual field difference
disappeared in Experiment 7. As has been mentioned before, the task difference found
Discussion
in Experiment 6 might have been due to a speed-accuracy trade-off. Therefore, there
may not be a genuine response time difference between the two tasks. The lack of an
overall visual field difference in Experiment 7 may reflect a shift of visual field
advantage across the four experimental sessions. This possibility will be discussed next.
Why is there no interaction between the task and the visual field in either
Experiment 6 or 7? It would appear that the position of the picture is not crucial for
producing the interaction since neither Experiment 6 (centered picture) nor Experiment
7 (lateralized picture) produced the task-visual field interaction. A factor which might
reduce the possibility of demonstrating an interaction might be found in the
experimental procedure. In both experiments, I used an ABBA design, in which four
sessions of 64 trials of either the semantic task or the phonological task were
performed in one hour on the same day. There is a possibility that a previous task
affected a performance on the next task, which might, therefore, mask a true visual
field difference. To examine this hypothesis, visual field advantages of the two tasks
in each session were calculated and plotted in Figure 3.9.
The abscissa represents session number and the ordinate represents a visual field
advantage. The visual field advantage was calculated by subtracting a mean reaction
time of the LVF from that of the right visual in each task for each session. Thus,
positive numbers indicate a LVF (RH) advantage, while negative numbers indicate a
RVF (LH) advantage. Two bars in each session indicate two groups of subjects. The
left bar represents the group who started with the semantic task, while the right bar
indicates the group who started with the phonological task. Figure 3.9(a) depicts the
results from Experiment 6 and Figure 3.9(b) represents the results from Experiment 7.
68
The direction of advantage varied as can be seen in Figures 3.9(a) and (b). In
Experiment 6, there were slight LVF advantages across sessions, although the
Insert Figures 3.9(a) and (b)
magnitude of the advantage varied (See Figure 3.9(a)). In Experiment 7, visual field
advantages shifted across sessions (See Figure 3.9(b)). This shift of advantage across
sessions might explain why there were neither overall visual field differences nor an
interaction between visual field and task in Experiment 7. (On the contrary, there were
consistent LVF advantages regardless of task in Experiment 6, except for the semantic
task in session 2. )
The tendency for the magnitude of the advantage to be greater when the
pictures were presented to the same side as the words (Experiment 7) than when the
pictures were located in the center (Experiment 6) is consistent with the findings in
experiments in Chapter 2. Therefore, the overall visual field differences mentioned in
the results section were deceptive, since it seemed that Experiment 6 produced greater
visual field advantage than Experiment 7. As a matter of fact, Experiment 7 showed a
larger visual field difference, but these differences were in opposite directions across
sessions (see Figure 3.9'(a)(b) for a summary).
Klein at a1. (1976) showed that a previous LH oriented task (e. g. English word
recognition) influenced a performance of next RH oriented task (e. g. face recognition)
and vice versa. This suggests that the mixing of the two tasks in the current study
might have washed out any visual field differences. Therefore, the first session was
analyzed separately in order to examine the subjects' performance that is
69
uricontaminated by prior sessions. The next section focuses on the analyses of the first
sessions of Experiment I and 2.
ANOVA for Session I: A three way ANOVA with one between-subject and two
Insert Figure 3.9 (a)(b)
within-subject factors was performed on reaction times of correct responses for
Experiment 6 and 7. The task was the between-subject factor, since half the subjects
started with the semantic task and the other half started with the phonological task.
The within-subject factors were visual field and character type.
The ANOVA for Experiment 6 showed a main effect of character type
(F(3,90)=22.23, P<0,001), which is consistent with previous results. (See Figure 3.11(a).)
None of the other effects were significant. Although there was no significant
interaction between task and visual field (F(I, 30)<I), there was a tendency for a LVF
advantage in both semantic and phonological tasks. (See Figure 3.10(c).)
The ANOVA for Experiment 7 showed the same results as above. That is, there
was a main effect of character type (F(3,42)=17.47, p<0,001), but none of the other
effects were significant. However, the interaction between task and visual field
showed a nearly significant trend for a LVF (RH) advantage in the semantic task and
a RVF (LH) advantage in the phonological task (p=0.10). (See Figure 3.12(c).)
------ -----------------
--------------------------------------------------
Insert Figure 3.10 (a)(b)(c) and 3.11 (a)(b)(c)(d)
The analyses of first sessions of Experiment 6 and 7 seem to suggest that we
might find an interaction between task and visual field if we isolate the two tasks (i. e.
perform the two tasks on different days) and present both picture and character to the
same visual field. The next experiment examined this hypothesis.
70
Insert Figures 3.12 & 3.13
This experiment examined whether there was an interaction between task and
visual field when the semantic and phonological tasks were performed on different
days so that hemispheric activation or other influences from one task would be less
likely to affect performance on the other task. Pictures and Kanji words were
Experiment 8: Isolating the two tasks
\
presented to the same side of the visual field since a greater hemispheric asymmetry
was found when the pictures were lateralized (session I of Experiment 7) than when
they were centered (session I of Experiment 6).
Subjects: Subjects were 12 Japanese people (2 males and 10 females) who had
resided in the Boston area for about one and half year, ranging from I month to 4
years' They often spoke and read Japanese. The average age of the subjects was 25
ranging from 20 to 30 years of age. All subjects were right-handed without any
immediate left-handed family member, and had normal vision. The subjects were paid
$6.00 for two 30-minute sessions.
Method
Apparatus: The apparatus was the TV monitor, computer, and video camera
used in Experiment 2.
changes. The stimuli were divided into two sets, so that each set had eight pictures
and 32 corresponding Kanji characters, This was necessary because I wanted to make
sure that equal number of the four types of character appeared once in each visual
71
Stimuli: The stimuli were the same as in Experiment 2 except for minor
field in each session.
Procedure: The procedure was the same as in Experiment 7 except for two
changes. First, the semantic and phonological tasks were performed on different days,
one-week apart. Half the subjects performed the semantic task first, while the other
half carried out the phonological task first. Each task consisted of two sessions. Half
the subjects started with one set of stimuli first, and the other half started with the
other set of stimuli first. Thus, the order of the task and set of stimuli were
counterbalanced across subjects.
untilthe subject pressed the response keys with both hands as in Experiment 6.
Consequently, the reaction times of Experiment 8 was shorter than those of Experiment
Second, reaction times were measured from the onset of the Kanji character
7.
The durations of stimulus presentation and ISI were the same as in Experiment
7. Thus, the picture was presented to either the LVF or RVF for 240 msec, followed
by ISI of one second, and the Kanji character appeared for 150 msec in the same side
as the picture.
After each task was completed, the subject was asked to recall Kanji characters
they saw during the experiment and write down as many Kanji as they could
remember. This immediate recalltest was performed in order to examine whether the
subject remembered a particular type of character more than the other types of
character, perhaps depending on the task.
Reaction time: Reaction times were logarithmically transformed prior to the
analysis for the same reason as in Experiment 6 and 7.
72
responses with 3 within-subject factors: task (semantic vs. phonological), visual field
(right vs. left), and character type (standard, homophone, semantically related, and
A 3-way 2 x 2 x 4 ANOVA was carried out on reaction times of correct
unrelated).
The results showed a main effect of visual field (F(I, ll)=5.47, p<0.05). That is,
responses to the LVF were faster than responses to the RVF (870 msec and 887 msec,
Result
respectively). The main effect of character type was also significant (F(3,33)=19.31,
p<0,001). The Newman Keuls test indicted that responses to the standard characters
(787 msec) were faster than responses to the semantically related characters (942 msec)
and unrelated characters (914 msec). They were not statisticalIy faster than responses
to the homophones (878 msec). There was no significant main effect of task
(F(I, ll)=1.26, P>0.10), however.
More interestingIy, there was a significant interaction between task and visual
field (F(I, ll)=4.70, p=0.05). A planned comparison indicated that in the semantic task
responses to the LVF (839 msec) were faster than responses to the RVF (864 msec)
(t(11)=2.06, p<0.05 with one tailed test), whereas in the phonological task there was no
significant difference between the two visual fields (902 msec for the LVF and 909
msec for the RVF, t(11) < I, P>0.05). None of the other interactions were statisticalIy
significant. (See Figures 3.14 and 3.15. )
There was a tendency for semantic interference in the phonological task,
whereas phonological interference was not observed in the semantic task. (See Figure
3.15 (b).) Although the interaction between task and character was again not
statisticalIy significant (F(3,33)=1.73, P=0.18), this tendency was consistent across the
three experiments.
73
analysis. However, since the results of ANOVA for the error rates and arcsine error
rates were almost identical, the results of ANOVA based on the untransformed error
Error rates: The error rates were converted to arcsine error rates for the
------------------------------------
rates are discussed in this section.
A 3-way 2 x 2 x 4 ANOVA with 3 within-subject factors was employed for the
analysis. The 3 factors were task (semantic vs. phonological), visual field (right vs.
left), and character type (standard, homophone, semantically related and unrelated).
The ANOVA showed a significant main effect of character type (F(3,33)=12.53,
P < 0,001). That is, the homophones and semantically related characters yielded more
errors than the standard and unrelated characters (10%, 13%, 6%, and 2.7%). This
Insert Figures 3.14 & 3.15
result was consistent with the results from Experiment 6 and 7. There were no
significant main effects of task (F(I, ll)<I, P>0.05), or visual field (F(I, ll)<I, P>0.05).
In this case, there was ino speed-accuracy trade-off in visual field. The LVF advantage
in reaction time (17 msec faster) was genuine.
The ANOVA also showed a significant interaction between task and character
(F(3,33)=14.0, p<0,001). A post-hoc analysis indicated that in the semantic task the
semantically related characters yielded more errors (16%) than the other three types of
characters (3.8% on average) (p<0.05), while in the phonological task the homophones
produced more errors (16%) than the other three (6% on average) (p<0.05), as shown in
Figure 3.17(b). This tendency was the same in both visual field (See Figure 3.17(d)).
None of the other effects were statisticalIy significant. Since there were virtually no
differences in error rates between the two visual fields in either task (F(I, ll)<I,
P>0.05), the significant interaction between task and visual field found in reaction
times was also genuine (see Figure 3.16(c).)
74
in this section. The standard characters were recalled the most and the unrelated
Immediate recall score: The results of the immediate recallscores are described
characters were recalled the least (12.2 and 2.2, respectively, with a maximum score of
16). The homophones and semantically related words were recalled almost as often (6.3
and 6.8, respectively). This pattern is similar for both the semantic and phonological
tasks. However, there was a suggestion of an interaction between the task and the
character type. That is, the semantically related characters were recalled slightly more
often than the homophones in the semantic task (7.2 and 5.4 respectively). The
homophones were recalled slightly more often than the semantically related characters
in the phonological task (7.2 and 6.3, respectively).
---------------------------------------
Insert Figures 3.16 & 3.17
The results from Experiment 8 showed that responses to the LVF (RH) were
faster than responses to the RVF (LH) in the semantic task, while there was no visual
field differences in the phonological task. In general, stimuli presented in the LVF
were responded to faster than the those presented in the RVF. This overall LVF
advantage seems to be mainly due to the greater LVF advantage in the semantic task.
performed on different days, they did not influence each other. Therefore, the
capacity of the RH to respond faster to Kanji characters in the semantic task could be
Discussion
These results can be explained as follows. Because the two tasks were
expressed.
A slight but non-significant LVF advantage was also observed in the
phonological task. This may be due to the nature of the picture-word matching task.
Thus, the pictures might have activated the RH more and biased the processing of the
Kanji characters towards the RH, even in the phonological task which normally
produce a LH advantage. 2
Another possible explanation is that visual-semantic processing of Kanji
characters is so strong that the RH activity overrides the LH involvement needed for
phonological processing. This tendency might be greater especially when the picture-
word matching task is performed since the task may not require overt phonological
processing. If I assume that the semantic processing for familiar and concrete Kanji
words takes place in the RH, and, additionally, that the picture-word matching task
minimizes overt phonological processing, any left-hemisphere advantage due to
2. However, this explanation can only be tentative since Experiment I in the currentstudy revealed that Kana words generated a LH advantage even in a semantic taskthat did not require overt phonological processing. If the picture-word matchingtask itself is responsible for producing the RH advantage, the LH advantage forKana words should also be weakened in Experiment I. The Kana words used in allthe experiments were visually unfamiliar, and may therefore have been processedphonologicalIy. In this case, the strong phonological aspect of Kana words mighthave overridden any RH activation.
75
phonological processing would be reduced. As a result, the overall visual field
differences in the phonological task might have been canceled out.
Consistently across the three experiments, semantic interference was observed in
the phonological task, while phonological interference was not observed in the semantic
task. (See Figures 3.2(b), 3.6(b) and 3.15(b).) In other words, similarity in meaning
interfered with sound judgment of Kanji characters, but similarity in sound did not
interfere with judgment of meaning. This supports the idea that Kanjiis processed in
the order grapheme-morpheme-phoneme. The results may indicate that grapheme-
morpheme conversion contributes most to the overall processing time. If so, processing
Kanji would require more RH activity in general. The RH involvement would be
expressed more clearly in the semantic task, whereas it would be balanced out or
reduced by the LH activity required for phonological processing in the phonological
task. This could account for the task-visual field interaction obtained in Experiment 8.
revealed different directions of hemispheric advantage, I examined this possibility in
Experiment 8. Hemispheric advantages in each session in each task are shown in
Figure 3.18. The positive numbers on the ordinate indicate the LVF(RH) advantage
while the negative numbers indicate the RVF(LH) advantage. The bars on the left in
each section represent the group who started with the semantic task first, whereas the
bars on the right in each section represent the group who started the phonological task
Since the breakdown of the tasks into four sessions in Experiment 6 and 7
76
first.
The figure shows that in the first session there was a great RH advantage (60
msec) in the semantic task, while there was only a slight (insignificant) RH advantage
(8 msec) in the phonological task. The other sessions showed a consistent RH
advantage, except for session 2 in the semantic task and session 4 in the phonological
task.
When the subjects started with the semantic task, there was a great drop of the
RH advantage from session I (60 msec) to session 2 (3 msec) in the semantic task. Four
of the six subjects showed this tendency. The same subjects showed a shift of
hemispheric advantage from the right to left in the phonological task (25 msec and -14
msec, respectively). Four of the six subjects also showed this tendency. However,
when the subjects started with the phonological task, they showed a small but
consistent RH advantage in the phonological task (8 msec for session I and 19 msec for
session 2). The RH advantage remained in the semantic task for these subjects (26
msec for session 3 and 18 msec for session 4) (see Figure 3.18).
These results are interesting but difficult to explain at this point. The
literature suggests shifts in hemispheric advantage due to familiarization and practice
(e. g. Hellige, 1976; Kittler, Turkewitz and Goldberg, 1989; Kosslyn at a1. , 1989). As
session I of each experiment showed greater hemispheric effects and was not
contaminated by any previous task, the results from session I were assumed to reflect
an optimal performance or strategy.
Session I of Experiment 2 and 3: Therefore, I examined whether combining the
results from session I of Experiment 2 and 3 would yield a significant interaction
between task and visual field in the next section. Although there is a minor difference
in stimulus distributions in session I between Experiment 7 and Experiment 8, the
results from sessions I of the two experiments were combined under the following
rationale. The proportion of each character type appearing in each visual field in
session I of Experiment 7 was about 0.25, ranging from 0.20 to 0.28 (s. d. =0.02). The
proportions of LVF trials and RVF trials were 0.53 and 0.47. Therefore, it is assumed
77
that each character type was distributed evenly enough in each visual field in
Experiment 7. In Experiment 8, the number of each character type in each visual field
was equal in each session because the experiment was designed this way.
A 3-way ANOVA with one between-subject factor and two within-subject
factors was performed on correct response times for session I, combining Experiment 7
and 8. There were 28 subjects. The between-subject factor was task, and the within-
subject factors were visual field and character type. The ANOVA showed that there
was a main effect of character type (F(3,78)=21.30, p<0,001), a marginal or non-existent
effect of visual field (F(I, 26)=3.29, P=0.08), and 00 main effect of task (F(I, 26)<I).
More interestingIy, the interaction between visual field and task was significant
(F(I, 26)=7.22, P=0.01). A planned comparison revealed that in the semantic task, the
stimuliin the LVF(RH) were responded to faster than those in the RVF(LH),
(t(27)=2,2588, p<0.05 with a two tailed test) 1405 msec and 1469 msec, respectively. In
the phonological task there was no difference between the LVF and the RVF, (t(27)<I)
(1439 msec and 1428 msec, respectively). This interaction is depicted in Figure 3.19.
None of the other interactions were significant.
The results from the combined session I support the prediction of an interaction
between task and visual field, but only when the picture and word are presented to the
same visual field and the tasks are separated in time.
Insert Figure 3.18 & 3.19
78
The series of experiments in this study showed three interesting findings.
Firstly, in the semantic task Kanji characters were processed faster in the RH than in
the LH, while in the phonological task there was no difference in processing time
between the two hemispheres. However, this task-hemisphere interaction was observed
only when the two tasks were isolated and performed on different days (Experiment 8
or session I of Experiment 7). When the tasks were carried out successively in ABBA
design on the same day, the interaction was not observed (Experiments 6 and 7).
A breakdown of each task into sessions revealed that in Experiment 7 the first
General Discussion
session showed a tendency of task-hemisphere interaction, but the later sessions showed
inconsistent directions of advantage. When the last three sessions were combined, the
direction of advantage was opposite to those of the first session (see Figure 3.9'(b)).
This opposite direction of advantage across sessions seems to cancel the initial tendency
of the interaction between task and visual field. This result leads to the hypothesis
that when the two tasks require different hemispheric functions and are performed one
after another, residual effects of one task might influence the performance of the
other task, perhaps due to confusion of the tasks or due to priming one hemisphere by
one task. Therefore, overall hemispheric-task interaction might disappear.
The results combining the first sessions of Experiment 7 and 8 showed that in
the semantic task the RH processed the Kanji characters faster than the LH, while in
the phonological task no hemispheric difference was observed (see Figure 3.19). These
resemantic task. the highly familiar and highly concrete Kanji characters were
processed better in the RH in the semantic comprehension task. The results replicated
the results from the experiments in Chapter 2 when only the standard and unrelated
characters were used.
However, in the phonological task, no hemispheric difference was obtained,
although a LH advantage had been expected. One interpretation of this finding is that
the visual-semantic association of Kanjiis so strong that RH is still involved in
processing the Kanji characters even in the phonological task. A lesser involvement of
the LH may be due to the nature of the picture-word matching task. This task reduced
overt phonological processing successfully, so that the LH involvement due to
phonology might have been minimal. An alternative interpretation is that the use of
pictures might have activated the RH, so that the LH function of matching sounds
competed with the RH function. As a result, the LH advantage for phonological
79
processing might have been canceled out.
A second important result was the tendency for the magnitude of hemispheric
advantage to be greater when the pictures were presented to the same side as the Kanji
characters as opposed to the center (See Figure 3.9 (a) & (b)). This result replicated the
results from the experiments in Chapter 2. These results seem to support the idea that
when pictures are clearly represented in both hemispheres, the two hemispheres are
activated equally well, so that hemispheric advantages are not easily observed.
However, when the pictures are presented peripheralIy for a short period, the visual
representations of the pictures are not perfectly clear. (Indeed, error rates were
somewhat higher (3%) in the side than the center (I%) conditions. ) The RH might be
rather heavily involved in processing these degraded pictures. Consequently, the
pictures activate the RH more than the left, and incoming Kanji characters, which
happen also to be processed better in the right hemisphere, may be then processed
more efficiently. The magnitude of hemispheric advantage should be maximized or
amplified in this condition. This might be one of the reasons why the phonological
task did not yield hemispheric difference. The normal, left hemispheric advantage of
phonological processing might have been canceled out by the increased activation of
the RH consequent on the degraded pictures.
If this is the case, why did the Kana words stillshow a LH advantage? One
possible explanation is that the phonological processing of the Kana words was so
strong that it might have overridden the RH activation by the pictures. This is
especially true in the experiments in Chapter 2 since alithe Kana words were
transcriptions of the Kanji words, and so were much less visually familiar; this is
claimed to require phonological processing (Besrier and Hildebrandt, 1987; Hirose, 1984,
1985). To test this hypothesis, one could use both unfamiliar and familiar Kana words
in the same task. If the familiar Kana words (i. e. words that are normally written in
Kana) are processed via direct visual code as suggested by Besrier and Hildebrandt
(1987), then the LH advantage caused by phonological processing should be reduced for
these Kana words. Consequently, I might find no hemispheric difference as I have
seen in the case of the Kanji characters, This hypothesis is interesting, but yet to be
80
tested.
Finally, I found a consistent tendency that semantic interference occurred in
judgments of sound whereas a phonological interference was not observed in judgments
of meaning. This supports the idea that access to the meaning precedes access to the
sound in the Kanji characters, so that the similarity of meaning confused the judgment
of sound but the similarity of sound did not interfere with the judgment of meaning.
As have been discussed earlier, this could also account for the task-visual field
interaction, since grapheme-morpheme conversion in Kanji might contribute most to the
overall processing time, and requires more RH involvement.
81
The purposes of this research were to examine whether Japanese Kanji and
Kana are processed in different hemispheres and to investigate if the right hemisphere
has some semantic processing ability for Kanji words. The results of this research
supported the idea that Kanji words are processed better in the right hemisphere (RH)
while Kana words are processed better in the left hemisphere (LH), in a picture
matching task. The results also supported the hypothesis that the RH is involved in the
semantic processing of Kanji words. The RH advantage in semantic processing of
Kanji words was inconsistent with the conclusions of some of the previous studies
(Hatta, 1979, 1981; Hayashi and Hatta, 1982) which showed LH advantages in their
semantic tasks. Picture-matching was assumed to be a sensitive task for examining the
comprehension of meaning, without involving either overt phonological processing or
higher cognitive functions, such as superordinate categorization. The use of a
lateralized picture presentation was also shown to maximize laterality effects.
In this chapter, I will summarize the important findings and then interpret
them. Finally, I willsuggest some directions for future research.
Chapter 4: General Discussion
82
Summary of Findings
I found five important results in this series of experiments. First, in the picture-
word matching task, Kanji words were processed faster in the RH while Kana words
were processed faster in the LH (Experiments I and 2). This result was complementary
to the findings of lones and Aoki(1988) who found that the stimulus duration needed
to achieve approximately 75 % accuracy was shorter in the RH for Kanji and shorter
in the LH for Kana. They also used the picture-word matching task.
Second, the laterality effect described above was obtained only when both the
picture and the word were presented to the same visual field (Experiment I and 2).
The laterality effect was reduced or disappeared when the picture was centered and
the word was lateralized (Experiment 3 and 4). This unforeseen result was interpreted
in Chapter 3 on the basis of Kirisbourne's activation theory (Kmsbourne, 1973, 1975).
Third, I found a task-visual field interaction in Kanji processing. That is,
Kanji words were processed faster in the RH in the semantic task but showed no
hemispheric difference in the phonological task (Experiment 8). The results of the
semantic task replicated the previous results (Experiment I and 2), which strongly
suggest that comprehension of Kanji words is better in the RH, at least when the words
are highly concrete and highly familiar. However, we do not know if unfamiliar and
abstract Kanji words would show the same pattern.
Fourth, the interaction between task and visual field for Kanji described above
was obtained in the AABB blocked design (i. e. Experiment 8), but not in the ABBA
design (i. e. Experiment 6 and 7). That is, when the two tasks were performed on
separate days, I found a RH advantage in the semantic task and no hemispheric
difference in the phonological task. However, when the order of the two tasks were
mixed on the same day, no laterality effect was found in either task.
Fifth, I found that there was semantic interference in the phonolo icaltask
while there was no phonological interference in the semantic task. Thus, semantic
information seems to interfere with phonological judgments, but phonological
information does not influence semantic judgments.
83
Interpretations of the Findings
I) Semantic Processing
The results of this research suggest that, for comprehension, Kanji words is
better in the RH while Kana words is better in the LH. The notion of a RH advantage
for Kanji and a LH advantage for Kana is not new, however. For exam Ie, K fj
words were reported more accurately in the RH and Kana words were reported more
accurately in the LH, in simple recognition tasks in which subjects had to name words
(Hatta, 1977; Hatta, 1978; Hirata & Osaka, 1967; Sasanuma, at a1. , 1977). The
recognition or naming tasks, however, do not necessarily require understanding of word
meaning but require phonological processing and access to the name. The findings of
the current study indicate that low-level semantic processing of Kanji and Kana words
also shows the same pattern of hemispheric advantage as the recognition or naming
tasks.
The results from the comparison of Kanji with Kana can be interpreted solely
in terms of differences in visual complexity and in the amount of phonological
processing. The RH advantage for Kanji may be purely due to the additional visual
processing load needed for the rather complex Kanji characters, while the LH
involvement for Kana might be purely due to the strong phonological processing
demand of Kana characters (Sasanuma at a1. (1980)). Therefore, the comparison
between Kanji and Kana does not itself prove that semantic processing occurs in the
RH.
There is, nevertheless, some evidence that supports the idea that the visual
complexity is not sufficient to account for the different hemispheric advantages
between Kanji and Kana. One bit of evidence comes from two studies with German
subjects in which physical matching of Kanji and Kana characters were required.
(Bussing at a1. , 1987; Hartje at a1. , 1986). The studies revealed that there was no
interaction between visual complexity of the two types of character and visual field.
That is, the German subjects, who were not familiar with Japanese written language,
did not show hemispheric advantages for either Kanji(simple and complex) or Kana in
performing a physical identification task. Thus, the results indicate that the
84
difference in visual complexity of two types of character does not introduce
hemispheric asymmetries in a purely perceptualtask. An analysis of complexity of
Kanji characters in the current study using the data of Experiment 6, also showed no
interaction between visual complexity and visual field in any of the tasks. (There is no
literature examining the visual complexity effect for Kanji using Japanese subjects,
however. )
More evidence comes from the experiments in Chapter 3 examining the
interaction between task and visual field for Kanji words. If visual complexity was
the only factor inducing the RH advantage for Kanji words, then both the semantic
and phonological tasks should have shown the same degree of RH advantage since the
same characters were used in both tasks. However, I found a task-hemisphere
interaction. Therefore, visual complexity alone cannot account for the results in this
study.
Then, what are the factors responsible for the interaction between task and
hemisphere? One factor is the RH's semantic processing capacity. If both semantic
and phonological processing took place only in the LH, then I should have found the
same pattern of hemispheric difference (i. e. a LH advantage) in both tasks since the
same stimuli were used. However, I found a right hemispheric advantage in the
semantic task and no hemispheric difference in the phonological task. This does
suggest that the RH has some semantic processing capacity.
The RH advantage for semantic processing of Kanji words, however, was
apparently inconsistent with previous findings in semantic tasks (Hatta, 1979, 1981a,
1981b; Hayashi & Hatta, 1982). These semantic tasks, however, involved higher
cognitive processing, such as superordinate categorization or judgment of congruency
between the meaning of the word and its location, rather than mere comprehension of
word meaning. Therefore, the results from the current study and the previous studies
suggest that lower level semantic processing of Kanji, such as comprehension of word
meaning, is still better in the RH, while higher level semantic processing, such as
superordinate categorization or semantic congruency, is better in the LH.
The RH advantage for semantic processing was obtained only for Kanji words
but not for Kana words. This is consistent with clinical data. Japanese aphasics and
85
dyslexics often show comprehension of Kanji words by pointing to pictures
corresponding to the words, while they often fail to do so for Kana words (see Faradis
at a1. , 1985; Sasanuma, 1980 for reviews). These patients usually have LH damage.
Therefore, damage has been done to either a phonological processing area of the LH or
a pathway from visual cortex to the phonological processing area or to a semantic
processing area. Since access to the lexicon in Kana usually requires phonological
processing (Gory0, 1987; Kimura, 1984; Sait0, 1981), the patients can neither read aloud
nor understand Kana words. In comparison, since access to the Kanjilexicon is
achieved via visual orthographic codes (Gory0, 1987; Kimura, 1984; Sait0, 1981), the
intact RH may be able to perform the semantic processing of Kanji words when post-
lexical phonological processing (i. e. reading aloud) is not required. (It would be
interesting to examine whether these patients can perform the superordinate categorical
task for Kanji words, since the task is assumed to be a LH function based on the
studies with normal subjects (e. g. Hatta, 1981; Hayashi & Hatta, 1982).) Taken
together, the results in this study and clinical observations confirm the idea that the
RH has some semantic capacity to comprehend Kanji words.
The semantic processing for Kanji words in the RH, however, may be limited to
highly familiar and highly concrete Kanji words. It has been suggested that in
tachistoscopic studies, concrete words are recognized relatively better than abstract
words in the RH due to their imageability, while abstract words are processed better in
the LH (Ellis & Shepherd, 1974; Hines, 1977). It has been also reported that some split-
brain patients could recognize concrete nouns in the RH (Gazzaniga, 1970; Zaidel and
Peters, 1983, Sperry, 1982). I intentionally used highly concrete, highly familiar and
highly hieroglyphic Kanji nouns in this study to establish an evidence that the RH has
some semantic processing capacity for these Kanji words. It would be interesting to
examine if the RH can process other types of Kanji words, such as abstract nouns,
verbs, unfamiliar words, etc.
86
There is an alternative view of Kanji lateralization based on clinical
observations. Iwata (1984) has suggested that the LH is responsible for reading both
Kanji and Kana, but the processing of Kanji and Kana involves different
intrahemispheric mechanisms. Based on observations of different types of Japanese
alexics and aphasics with damage to different locations in the LH, Iwata (1984)
proposed that visual information arriving at the left visual cortex is conveyed to
Wernicke's area by two different pathways in the LH. One is a dorsal pathway via the
angular gyrus, which serves mainly phonological processing for Kana reading. The
other is a ventral pathway through the posterior part of the middle and inferior
temporal gyri, which serves semantic processing for Kanji reading. The assumption of
different intrahemispheric mechanisms for Kanji and Kana is supported by several
cases of Japanese alexics (e. g. Kawamura at a1. , 1987; Mochizuki & Ohtom0, 1988) and
agraphics (e. g. Kawamura at a1. , 1989; Tanaka at a1. , 1987).
However, these clinical data do not necessarily exclude the possibilitity of
semantic processing of Kanji words in the RH. One reason is that the patients reported
in the literature were forced to read the words aloud, which is a function of the LH.
It is possible that the patients could understand the Kanji words, but were not able to
read them. In fact, Mochizuki and Ohtom0 (1988) suggested that Kanji words arriving
in the right visual cortex may be transferred to the right middle temporal gyrus via the
inferior longitudinal fasciculus, and further transferred to the left middle tern oral
gyrus from the right gyrus via the intact corpus callosum, and finally conveyed to
Wernicke's area. They proposed this mechanism based on the observation that their
pure alexic patient, who had lesions in the left occipital lobe and inferior temporal
gyrus, partially recovered his Kanji reading ability about I month after the onset of
his alexia. Kawamura at al. 's patient (1987) also could have understood the meaning of
Kanji words but could not read aloud perhaps due to destruction of the left posterior
inferior temporal gyrus, which might send the semantic information to Wernicke's area.
Consequently, these clinical data do not prove that both Kanji and Kana are processed
dominantly in the LH. More careful examinations seem to be necessarily in the clinical
87
studies.
2) Phonological Processing
The results in the current study showed that there was no hemispheric
difference in the phonological task for Kanji words. This result was inconsistent with
a previous study. Sasanuma at a1. (1980) showed a LH advantage for Kanji in their
phonological task. This discrepancy might be due to a task difference as well as
stimulus difference. I used a successive picture-matching task while Sasanuma at al.
employed a simultaneous word-word matching task. I used only highly familiar and
highly concrete Kanji nouns (except for some homophones) while they used high-
frequency Kanji characters, They did not specify other properties, such as concreteness
or word class (verb or noun). In the phonological task of this study, phonological
processing of Kanji words in the LH might have been canceled out by a RH
involvement. Such a RH involvement in this phonological task may be due to one or a
combination of the following factors: I) Kanji words are processed better in the RH
due to extra visual perceptual processing. 2) High familiarity and/or high concreteness
may enhance RH processing. 3) Picture processing prior to Kanji processing biases the
RH. 4) The picture-word matching task effectiveIy minimizes overt phonological
processing, so that the LH involvement is reduced.
88
3) Effect of Lateralizing the Picture
I found a larger interaction between orthography (Kanji vs. Kana) and visual
field when both the picture and the word were lateralized than when the picture was
centered and only the word was lateralized (Chapter 2). I also found that the
hemispheric advantage in each session tended to be larger in the lateralized picture
condition than in the centered picture condition (Chapter 3). These results were
interpreted according to Kmsbourne's activation theory (Kmsbourne, 1970,1973). When
the picture is lateralized the picture activates the hemisphere to which the picture is
presented. If an incoming stimulus requires processing by the activated hemisphere,
processing would be maximally efficient. However, if the incoming stimulus is
processed in the other hemisphere, enhancement does not occur.
The method of lateralized picture presentation, therefore, seems to maximize
subtle laterality effects.
4) Condition Order Effect
I found an interaction between task and visual field in Kanji words only in the
blocked design (AABB), not in the mixed design (ABBA). One can interpret this result
in two ways; in terms of activation or strategy. If activation were long-lasting, any
biased activation of one hemisphere caused by one task might influence performance
on the next task. Thus, if the two tasks favor different hemispheres, the activation of
one hemisphere provoked by one task might cancel a potential advantage of the other
hemisphere for performing the other task, when the two tasks were performed
successively as in the ABBA design. The other interpretation involves a strategy
effect, in which the strategy employed for one task is carried over in performing the
89
other task. For instance, one might recover the name of a character by using
phonological processing to access the lexicon in the phonological task, and then use the
same processing strategy to access the lexicon and recover the meaning of a character
in the semantic task. Both interpretations predict that the very first session shows
uriconfounded hemispheric effects, solely due to the task being performed.
This prediction was supported by an analysis of the advantage for each session
in each task. The first sessions of Experiment 7 and 8 revealed a strong RH advantage
in the semantic task and either a LH advantage or no advantage in the phonological
task. On the other hand, the later sessions showed shifts in or weakening of these
hemispheric advantages. Presumably subjects entered the experiments with neither
biased hemispheric activation nor biased strategy, so that the hemispheric advantage
was maximized in the first session. Klein at a1. (1976) obtained a similar finding.
They found that activating the LH with a verbal task reduced a RH advantage for
face recognition while activating the RH with a face-recognition task attenuated a LH
advantage for word recognition. However, at present, we do not know exactly how, or
how much, one task influences the performance of the other.
One discrepancy between this idea and the results in this study is that
Experiment 6 did not show the same pattern as Experiment 7 and 8. However, this
discrepancy can be accounted for by the picture activation hypothesis described in the
previous section. The hemispheric advantages obtained in each session in Experiment 6
were overall smaller than those in Experiment 7 and 8. The advantages in Experiment
6 might have been small due to centered picture presentation while those of the other
experiments might have been larger due to lateralized picture presentation. Since both
hemispheres were activated by the centered picture, and hence the hemispheric
advantage was small to begin with, any effect of condition order might have been to
90
small to be observed.
5) Semantic Interference
experiments: semantic interference in the phonological task, and no phonological
interference in the semantic task. This result is consistent with the idea that access to
The same pattern of interference was obtained throughout the three
the Kanjilexicon is not necessarily mediated by phonological processing. Instead, the
Kanjilexicon may be directly accessed via a visual code, and phonological processing
may occur later (Gory0, 1987; Sait0, 1981). Semantic similarity, therefore, may confuse
phonological judgment, while phonological similarity does not confuse semantic
judgment, perhaps because the latter is performed before phonological information is
processed. It would be interesting to investigate whether this finding is unique to
Kanji recognition or can be generalized to other orthographies, such as English.
Directions for Future Research
There are some questions yet to be answered regarding the semantic processing
capacity of the RH.
First, does this capacity generalize to all Kanji characters? For instance, can
abstract nouns and less familiar concrete nouns be comprehended better in the RH than
the left? Can other word classes of Kanji characters, such as verbs and adjectives, also
be comprehended better in the RH? Most of the clinical studies tested the
91
comprehension of concrete nouns, specifically using a picture-word matching task for
dyslexics and partial split brain patients (e. g. Sasanuma, 1980; Sugishita at a1. , 1978;
Sugishita at a1. , 1986; Sugishita and Yoshioka, 1987). The other word classes were
tested by oral reading (e. g. Sasanuma, 1980), or the word classes were not specified at
all(e. g. Iwata, 1984; Kawamura at a1. , 1987; Mochizuki and Ohtom0, 1988). Therefore,
it is still not clear that the semantic processing in the RH is only limited to highly
familiar and highly concrete Kanji words.
Second, are the results obtained in the current study due to familiarity,
concreteness or a combination of the two? One way to test this question is to hold the
one variable constant while varying the other.
Third, do the results generalize to two-character Kanji words? I used single
character Kanji words in this study. However, a two-character Kanji word is also
assumed to be a lexical unit in Japanese writing (Yokosawa, 1988). It has been
reported that two-character Kanji words are identified better in the LH (Hatta, 1978;
Tzeng at a1. , 1979). This result is usually interpreted as indicating that the two-
character Kanji words are processed sequentially and analyticalIy by the LH while
single-character Kanji words are processed holistically by the RH (Tzeng at a1. , 1979).
However, Hatta (1978) did not control for the concreteness of the Kanji words.
Moreover, the identification task involved oral reporting which is a LH function. It
might be still possible that high-frequency concrete two-character Kanji words are
comprehended better in the RH. One way to test this possibility is to use the same
experimental method as in the current study (i. e. the picture-word matching task with a
lateralized picture presentation) but with two-character Kanji words.
Fourth, can the RH understand highly familiar concrete Kana words? The
Kana words used in the current study were visually less familiar, because they were
transcribed from words that are normally written in Kanji. The visual unfamiliarity
of the Kana words precluded lexical access by an orthographic route, and might have
increased phonological processing requirements. Recent English language literature
suggests that lexical access to English words is often achieved without phonological
mediation when they are high-frequency words (e. g. Waters and Seidenberg, 1985).
Likewise, it has been reported that orthographically familiar Kana words (i. e. , loan
words written in Katakana) are named or categorized faster than visually less familiar
Kana words (i. e. ,transcriptions of Kanji words) (Besrier and Hildebrandt, 1987; Hirose,
1984, 1985). These results suggest that lexical access to visually familiar Kana words
can be achieved without phonological mediation. If a direct visual code is sufficient
92
.
for these Kana words, then it might be possible that these words can be comprehended
by the RH. On the other hand, if there is an intrinsic difference between Kanji and
Kana, then these Kana words should still be processed better in the LH. This
possibility is yet to be tested.
Fifth, can we obtain a RH advantage for high-frequency concrete English
words? Although an interaction between concreteness and visual field was obtained in
some studies (e. g. Ellis and Shepherd, 1974; Hines, 1976, 1977), overall RVF advantages
were generally reported. This might be partly due to extra phonological processing
required by naming tasks. If a visual orthographic code is sufficient to access high-
frequency words (especially concrete and imageable nouns), then we might obtain a RH
advantage for these English words in a picture-matching task.
Sixth, did a picture presentation create the extra RH involvement in the picture-
word matching task? To examine this possibility, a word-word matching task can be
utilized to eliminate picture processing. One can perform a semantic and phonological
task employing a target Kanji word and a test Kanji word. In the semantic task, the
test word can be either a word semantically similar to the target or a semantically
unrelated word. In the phonological task, the test word can be either a homophone of
the target word or a none-homophone.
93
electrophysiological recordings? Very few studies have been performed on the issue of
Japanese written language lateralization using electrophysiological recordings (e. g.
Hatta et. a1. , 1983; Hink at a1. , 1980). My findings would be more convincing if I could
obtain the same results from both behavioral measures (e. g. reaction times, error rates)
and electrophysiological measures (e. g. evoked potential), which may be lateralized
using appropriate electrode arrays (Srebr0, 1985a, b).
Seventh, can we obtain the laterality effects of Kanji and Kana using
The results of a series of experiments suggest a strong possibility that the right
hemisphere can comprehend highly familiar, highly concrete and highly hieroglyphic
single Kanji characters, This is a new finding since previous Japanese research has
shown that visual processing for Kanji (e. g. visual matching) is better in the right
hemisphere, but that semantic processing for Kanji(e. g. superordinate categorization) is
better in the left hemisphere. However, there are intermediate levels of processing that
this prior research did not address. Visual matching does not necessarily require
comprehension of Kanji words and semantic processing, as previously examined,
required higher cognitive processing, such as semantic categorization or judgment of
semantic congruency, than mere understanding of the word meaning. The current
study filled in the gap between the previous findings; the comprehension of meaning
of familiar and concrete Kanji words can be achieved in the right hemisphere.
Conclusion
94
MEAN REACTION TIME (r'sEc)
(c)(a)Exp I: LATERALIZED PICTURE Exp 3: CENTERED PICTURE
(150 HSEC) (150 MsEC)
RVF LVF LVFRVF
TABLE 2.1
KANJI 831
(154)
KANA 841
(194)
a I I
(143)
836
95
894
(188)
(b)Exp 2: LATERALIZED PICTURE
(500 MsEC)
RVF LVF
821
852
867
KANJI
879
(179)
KANA
863
(142)
861
(144)
877
(202)
908
(157)
870
827
(156)
906
(, 800
878
(d)Exp 4: CENTERED PICTURE
(I SEC)
RVF LVF
886
940
(154)
845
892
884
884
924
897
(263)
686
(136)
863
(242)
( ) indicates Standard deviation
89 I
883
(140)
880
873
884
(a)Exp I: LATERALIZED PICTURE
(150 MsEC)
RVF LVF
TABLE 2.2
KANJI
ERROR RATES (%)
KANA
4.3
(2.4)
4.2
(2.7)
2.6
(2.5)
(c)Exp 3: CENTERED PICTURE
(150 MsEC)
LVFRVF
4.25
(b)Exp 2: LATERALIZED PICTURE
(500 risEc)
RVF LVF
2.9
(2.2)
3.5
96
2.8
3.6
KANJI
1.3
( 1.7)
KANA
1.9
(2.2)
I . I
(2.0)
1.3
(2.4)
3.0
(33)
1.2
2.4
(3.4)
2.0
(2.00
1.3
(d)Exp 4: CENTERED PICTURE
(I SEC)
RVF LVF
2.5
3.2
(3.2)
2.2
1.7
1.6
3.1
2.8
I. 5
( 1.7)
0.6
( 1.2)
1.7
(2.0)
I . I
0.4
( 1.0)
) indicates Standard deviation
1.6
0.5
I . I
Figure 1.1 (a) An example of Kanji(logographic characters).
(b) An example of Kana (syllabic characters).
Figure 1.2 Some examples of homophones of Jinl listed by Iwata (1984).Figure 1.3 Some examples of Kanji characters with both phonological and semantic
elements (adapted from Sasanuma (1980)).
Figure 2.1 Character-visual field interactions in reaction times. (a) Lateralized picture
presentation for 150 msec in Exp. I to < 0.05). (b) Lateralized picture presentation for
500 msec in Exp. 2 to < 0.05). (c) Centered picture presentation for 150 msec in Exp. 3
Figure Legends
to < 0.05). (d) Centered picture presentation for I sec in Exp. 4 to < 0.05).
Figure 2.2 Individual data of hemispheric advantage for Kanji and Kana. Positive
numbers indicate RH advantages while negative numbers indicate LH advantages. (a)
Exp. I (b) Exp. 2. (c) Exp. 3. (4) Exp. 4.
Figure 2.3 Character-visual field interactions in error rates. (a) Lateralized picture
presentation for 150 msec in Exp. I to >0.05). (b) Lateralized picture presentation for
500 msec in Exp. 2 (p > 0.05). (c) Centered picture presentation for 150 msec in Exp. 3
to > 0.05). (d) Centered picture presentation for I sec in Exp. 4 to > 0.05).
97
Figure 2.4 Mean stimulus durations of 70 trials in the LVF and RVF in picture
recognition (p > 0.05).
Figure 2.5 Mean error rates in the LVF and RVF in picture recognition to > 0.05).
Figure 2.6 Mean stimulus durations of last 40 trials in the LVF and RVF in picture
recognition to > 0.05).
Figure 2.7 Mean stimulus durations of last 20 trials in the LVF and RVF in picture
recognition ( p > 0.05).
Figure 2.8 Visual field advantage for Kanji and Kana. (a) Lateralized picture
presentation (Exp. I & 2). (b) Centered picture presentation (Exp. 3 & 4).
Figure 3.0 Examples of four types of stimuli in the semantic and phonological tasks in
Exp. 6, 7 & 8.
Figure 3.1 Results of reaction times in Exp. 6. (a) A main effect of task to < 0,001). (b)
A main effect of visual field to < 0.05). (c) No task-visual field interaction to > 0.05).
Figure 3.2 Results of reaction times in Exp. 6. (a) A main effect of character type to <
0,001). (b) A task-character interaction to < 0,001). (c) No character-visual field
interaction to > 0.05). (d) No task-character-visual field interaction to > 0.05).
Figure 3.3 Results of error rates in Exp. 6. (a) A main effect of task to < 0,001). (b) No
main effect of visual field to > 0.05). (c) No task-visual field interaction to > 0.05).
Figure 3.4 Results of error rates in Exp. 6. (a) A main effect of character type to <
0,001). (b) A task-character interaction to < 0,001). (c) No character-visual field
interaction to > 0.05). (d) No task-character-visual field interaction to > 0.05).
Figure 3.5 Results of reaction times in Exp. 7. (a) No main effect of task to > 0.05). (b)
No main effect of visual field to > 0.05). (c) No task-visual field interaction to > 0.05).
Figure 3.6 Results of reaction times in Exp. 7. (a) A main effect of character type to <
0,001). (b) A task-character interaction to < 0.01). (c) No character-visual field
interaction to > 0.05). (d) No task-character-visual field interaction to > 0.05).
Figure 3.7 Results of error rates in Exp. 7. (a) A main effect of task to < 0,001). (b) No
main effect of visual field to > 0.05). (c) No task-visual field interaction to > 0.05).
Figure 3.8 Results of error rates in Exp. 7. (a) A main effect of character type to <
0,001). (b) No task-character interaction (p > 0.05). (c) No character-visual field
interaction to > 0.05). (d) A task-character-visual field interaction to < 0.05).
Figure 3.9 Visual field advantages for the semantic and phonological tasks in each
session. (a) Centered picture presentation for I sec in Exp. 6. (b) Lateralized picture
presentation for 240 msec in Exp. 7.
Figure 3.9' Visual field advantages for the semantic and phonological tasks in session I
and subsequent sessions combined. (a) Centered picture presentation for I sec in Exp. 6.
(b) lateralized picture presentation for 240 msec in Exp. 7.
Figure 3.10 Results of reaction times in first session of Exp. 6. (a) No main effect of
98
task to > 0.05). (b) No main effect of visual field to > 0.05). (c) No task-visual field
interaction to > 0.05).
Figure 3.11 Results of reaction times in first session of Exp. 6. (a) A main effect of
character type to < 0,001). (b) No task-character interaction to < 0.01). (c) No character-
visual field interaction to > 0.05). (d) No task-character-visual field interaction to >
0.05).
Figure 3.12 Results of reaction times in first session of Exp. 7. (a) No main effect of
task to > 0.05). (b) No main effect of visual field to > 0.05). (c) No task-visual field
interaction to = 0.10).
Figure 3.13 Results of reaction times in first session of Exp. 7. (a) A main effect of
character type to < 0,001). (b) No task-character interaction to > 0.05). (c) No character-
visual field interaction (p > 0.05). (d) No task-character-visual field interaction to >
0.05).
Figure 3.14 Results of reaction times in Exp. 8. (a) No main effect of task to > 0.05).
(b) A main effect of visual field to < 0.05). (c) A task-visual field interaction to <
0.05).
Figure 3.15 Results of reaction times in Exp. 8. (a) A main effect of character type to <
0.05). (b) No task-character interaction (p > 0.05). (c) No character-visual field
interaction to > 0.05). (d) No task-character-visual field interaction to > 0.05).
Figure 3.16 Results of error rates in Exp. 8. (a) No main effect of task to > 0.05). (b)
No main effect of visual field (p > 0.05). (c) No task-visual field interaction to > 0.05).
Figure 3.17 Results of error rates in Exp. 7. (a) A main effect of character type to <
0,001). (b) No task-character interaction to < 0,001). (c) No character-visual field
interaction to > 0.05). (d) No task-character-visual field interaction to > 0.05).
Figure 3.18 Visual field advantages in each session in Exp. 8 where the semantic and
phonological tasks were performed on separate days.
Figure 3.19 A task-visual field interaction for pooled data of the first sessions of Exp.
7 and 8 to < 0.05).
99
KANAKANJI
(LOGOGRAPH) (SYLLABARY)
^ I IMeaning
pronunciation: [kawa][sen]
100
River
^)^\ *")River
Skin
Ikawal
(a )
Figure I. I
( b )
Some Examples orhomophonesor Innllisted by Iwate ( I 984)
^^: :invade
;^; :sleep.^I
,~
^I'"^ :diagnose
^.^
: soak
: truth
: new
101
semantic phonologicalelements elements
Examples of Kanjicharaoterswith bothphonological andsemanticelements
(adapted from 56^numa (1980))
; needle
speech ^-,;^..j6^ ^Z
^:^..
I@
^ ^,..
Igol
Figure 1.2
word
[ hj9]
semantic phonologicalelements elements
criticism
[g9]
tree ^'.;C.
^:;^,
^:^..error
IC^,l
[shi]
cane
I heril
twig
board
Figure 1.3
(msec)950
Exp I: LATERALizED PICTURE
(, 50 MsEC)
LLl=I-.ZoI-Q<IU.=
900
850
800
^
^.^
(msec)
950
RVF
KANJI
KANA
N=, 2
VISUAL FIELD
(msec)
950
Exp 2: LATERALizED PICTURE
(500 MsEC)
LLl=,-ZoF=Q<IU.=
Exp 3: CENTERED PICTURE
(, 50 MsEc)
900
I"=- 900
=oI. .o<IU.=
LVF
2.1 (a)
^
850
102
850
800
800
RVF
N=12
(msec)
950
RVF
N=12
VISUAL FIELD
Exp 4; CENTERED PICTURE
(I SEC)
VISUAL FIELD
IL^=^ 900
=oI-o<Iu 850.=
LVF
2.1 (by
LVF
2.1 to)
800
Figure 2. I
N=, 2
F^/F
VISUAL FIELD
LVl=
2.1 (d)
-.,.,- . ~ .~'...-" . , ,~,'"~.. .. -'p, ..- ^.-,.~-,','.*".
(msec)
200
Expt: LATERALizED PICTURE
(150 MsEC)
^
LL>.
LL>.=
<Z<><
.
100
^
o
-100
-200
-200
. .
-100 o
KANJI(RVF. LVF)
(msec)200
..
.
Exp 2: LATERALizED PICTURE
(500 MsEC)
(msec)
200
^
.
LL>.
L.>.=
<Z<><
.
.
100
Exp3: CENTERED PICTURE
(, 50 MsEc)
^
IL>.
IL>^
<Z<a, -100
^
100
.
I 00
2.2 (a)
o
^
103
doo
o
200
(msec)
-200
-200
-200
.
-200
.
. .
.
.
.
-100 o
KANJI(RVF. LVF)
.
..
.
.
o0
.
-100 o
KANJI(RVF-LVF)
.
(msec)200
. .
.
Exp 4: CENTERED PICTURE
(, SEC)
^
L>.
LL>^
<Z<><
100
.
I 00
2.2 ^)
^
I 00
2.2 to)
o
200
(msec)
-100
200
(msec)
-200
.200
co
Figure 2.2
.
~
o.too
KANJI(RVF-LVF)
.
.
100
2.2 (d)
200
(msec)
(%)
5
Exp I: LATERALizED PICTURE
(, 50 MsEC)
--0- KANJI
11.1I-<C=
DC02.=P=IP
4
3
o
RVF
VISUAL FIELD
KANA
N. 12
(9, .)
5
Exp 2: LATERALizED PICTURE
(500 MsEC)
(%)5
11.1I-<C=
ECoa=DCLU
4
'Exp3: CENTERED PICTURE
(150 MsEC)
LUI. ..<P=
P=on=PCLU
3
4
LVF
104
2
a
2.3 (a)
2
o
o
N=12
RVF
VISUAL FIELD
RVF
VISUAL FIELD
(v^)
5
Exp 4: CENTERED PICTURE
(, SEC)
LUI-<P=
.=o.=P=LU
4
LVF
3
LVF
2
2.3 (by
2.3 (c)
o
Fi. gure 2 . 3
N=12
RVF
VISUAL FIELD
LVF
2.3 (d)
(msec)
150
Zo
Exp5: PICTURE RECOGNITION
(70 TRIALS)
<n==a
co=.==,-co
100
^
50
o
105
RVF
VISUAL FIELD
(%)25
^.
^
Exp5: PICTURE REcoGNTiON
co1.1
n=.=oP=^I"
<
LVF
20
HIGH INrENsrrY
Low INrENsrrY
15
10
Figure 2.4
5
o
RVF
VISUAL FIELD
^
^.-
LVF
HIGH INrENsrnr
Low INrENsrrY
Figure 2.5
(msec)
150
Zo
Exp 5: PICTURE RECOGNITION
(LAST40 TRIALS)
<.==Q
co=.==I-co
100
^
50
o
106
RVF
VISUAL FIELD
(msec)
150
Zo
^..-
^
<
Exp5: PICTURE RECOGNITION
(LAST 20 TRIALS)
P==a
co=.==I. .co
LVF
too
HIGH INrENsrrY
Low INrENsrn/
^
50
Figure 2.6
o
RVF
VISUAL FIELD
^--
^-.
LVF
HIGH INrENsrrY
Low INrENsrrY
Figure 2.7
(msec)200
LATERALIZED PICTURE PRESENTATION
^
LL>.
U.
>P=
<Z<><
.
100
^
o
-100
-200
-200
.
.
.
,,
,
.,
-100
(msec)
200
,,
,
107
.
,.
,
^~
LL
>.
LL
>P=
<Z<^
KANJI(RVF-LVF)
CENTERED PICTURE PRESENTATION
.
.
o
.
too
^
. Expi:SHORTDURATioN
, Exp 2:LONGDURATjoN
I 00
o
doo
200 (msec)
Figure 2.8 (a)
-200
.
.
,,,,
-200
.
. .
,
,,
.
-, 00
,.
..
.
.
KANJI(RVF-LVF)
,
o
. Exp3:SECRTDURATioN
, Exp4:LONGDURAT'10N
100 200 (msec)
Figure 2.8 (by
^
MEANING :
PRONUNCIATION :
FLOWER
IRANAi
++.
TASK/ TYPE
4 IC^FLOWER
IHA"A1
SEMANTIC
'^I:^.
PHONOLOGICAL
STANDARD
.^:.NOSE
IHA"A1
108
*
YES
HurloPHo"E SEMANTIC
GRASS
IKUSAi
,, L. \
1:11^'
YES
NO
ROCK
11WAl
PICTURE
YES
YES
UNRELATED
I SEC
OR
240 MsEc
KANJI
NO
NO
I SEC 150 MsEc
RESPONSE
NO
Figure 3 .O
REACTION
TIME
(msec)1000
LLl
=I-
ZoI-<9<1.1P=
^
900
^
Exp6:TASK
800
700
(msec)1000
SEMANTIC
U. I=I-
Zor-C. ><LLlP=
TASK
Exp 6: VISUAL FIELD
^
900
^
ProhCLOGICAL
109
800
700
Figure 3.1(a)
(msec)1000
LLl
=I. .
=oI. .<.><1.1P=
RVF
VISUAL FIELD
^
Exp 6: TASKvs. VISUAL FIELD
900
^
800
LVF
700
Figure 3.1(b)
RVF
VISUAL FIELD
-^,.^
^--
LVF
SEMANTIC
PFO\PLCGICAL
Figure 3.1(c)
(msec)
1000
LU=I-
ZoI.o<LU.=
900
Exp6: CHARACTER
800
700
600
STANDARD
(msec)
1000
SEMEuro
CHARACTERTYPE
LU
=I-
ZoI-o<1.1.=
Exp6:TASKvs. CHARACTER
900
(msec) Exp 6: CHARACTER vs. VISUAL FIELD
800
L, *^IATED
1000
700
I"=I. .
ZoI-o<IUPC
^
Figure 3.2 (a)
900
600
^
800
STANDARD
700
FunO SEM
CHARACTERTYPE
600STANDARD
^--
^
.
UNRELATED
(msec)
1000
Euro SEM
CHARACTER
SEMANTIC
R. ^LOOK;AL
LLl
=I-
ZoI-o<LUPC
Exp6:TASKvs. vFvs. CHARACTER
Figure 3.2 (by
900
^.-
^--
UNRELATED
800
I^IF
LVF
700
Figure 3.2 (c)
II,,
I,
600
I~
,,
IIIIIII
I
I
III
STANDARD
JP
~~
Euro SEM
CHARACTERTYPE
^
--.--
^
--.--
UNRELATED
SEMRVF
SEMLVF
PHONRVF
PHONLVF
Figure 3.2 (d)
I-.
F-,
o
(^^)
25
LLlI. -
^
20
.=on=n=LU
Exp6:TASK
15
10
5
o
(^
25
SEMANTIC
LLl
TASK
Exp 6: VISUAL FIELD
I-
^
20
n=on=.=Ll
15
111
10
ProneLOGICAL
5
o
Figure 3.3 (a)
(%)
25
VISUAL FIELD
Exp 6: TASKvs. VISUAL FIELD
RVF
LLlI-
^
20
.=on=.=LLl
15
10
LVF
5
o
Figure 3.3 (by
RVF
VISUAL FIELD
^.^
^.^
LVF
SEMANTIC
ProroLOGICAL
Figure 3.3 (c)
(%)
25
I"
.=
.=o.=.=1.1
I. .<
20
Exp6: CHARACTER
15
10
5
oSTANDARD
(%)
25
1.1C^ SEM
CHARACrERTYPE
Exp6:TASKvs. CHARACTER
I"
PC
.=o.=PCIU
r-<
20
15
Cfo)
25
UNRELATED
10
Exp6: vFvs. CHARACTER
LU
P=
.=o.=PCLU
I. .<
5
20
Figure 3.4 (a)
o
15
STANDARD
to
^.^
^
5
Fit^to SEM
CHARAerERTYPE
,III
A-
R. ^G
SEMANTIC
o
srANDARD
..
(%)
25
\
INRELATED
\
^a-^
--.-.
1.1C^to SEM
CHARAerERTYPE
Exp6:TASKvs. VF vs. CHARACTER
LU
.=
.=oPCPCLLl
<
20
RVF
LVF
Figure 3.4 (by
15
UNRELATED
10
,,,
,
5
I,,,,
Figure 3.4 (c)
,
o
\\
---,
\
STANDARD
\
,\
II
,I
^.-
--.-.
^.
--.-.
HOMO SEM
CHARACTYPE
\\
SEMRVF
SEMLVF
priorroRVF
ProneLVF
..
UNRELATED Figure 3.4 (d)
I-.
I. -.
I\>
(msec)
1900
in
=I-Zo
C. ><LLlP=
1800
^
Exp7:TASK
1700
1600
(msec)
1900
SEMANTIC
LU
=I-Zo,.C. .><IUP=
TASK
Exp 7: VISUAL FIELD
1800
^
ProneLOGICAL
1700
113
1600
Figure 3.5 (a)
(msec)
1900
RVF
Exp7: TASKvs. VISUAL FIELD
in
==I-Zo
<LUP=
VISUAL FIELD
1800
^
<.>1700
LVF
1600
Figure 3.5(b)
RVF
VISUAL FIELD
^
^
LVF
SEMANTIC
Pro\10LOGICAL
Figure 3.5(c)
(msec)
1900
LU=I-Zo
<LU.=
1800
^
Exp7: CHARACTER
^
1700
5 1600
1500
1400srANDARD
(msec)
1900
1.1^ro SEM
CHARAGrERTYPE
LU=I.
Zo
<IUPC
1800
^
Exp7:TASKvsCHARACTER
1700
^
F.,.>
(msec)
1900
1600
UNRELATED
Exp7: CHARACTERvs. VISUAL FIELD
1500
LU=,..Zo
<IUPC
1800
^
1400
Figure 3.6 (a)
1700
^
CD 1600
STANDARD
1500
^.
^.^
1.1(:^to SEM
CHARACTERTYPE
1400
,,
,,
,,
,
SEMANTIC
PHONG. 001CAL
,PO'~~~~~
srANDARD
(msec)
1900
UNRELATED
I-K^to SEM
CHARAerERTYPE
Exp7:TASKvs. vFvs. CHARACTER
1.1=I-
==o
<IU.=
^
--*--
1800
^
Figure 3.6 (by
1700
^
<.>
Rl!F
LVF
1600
UNRELATED
1500
Figure 3.6 (c)
1400
,II,I
II
~ ,,
II
,,
II
SFANDARD
~ ^
--a. ^
--.--
^.
--.--
F10rro SEM
CHARAcrERTYPE
SEMRVF
SEM LVF
pHORVF
pHOLVF
UNRELATED Figure 3.6 (d)
I-..
F-,
4=~
(%)
25
LLlI-
^
20
.=o.=.=LU
Exp7:TASK
15
10
5
o
(%)
25
SEMANTIC
LUI-
20
TASK
Exp7: VISUAL FIELD
.=o.=.=LLl
15
115
10
ProneLOGICAL
5
o
Figure 3.7 (a)
(%)25
RVF
LLlI. .
^
VISUAL FIELD
Exp7: TASKvs. vF
20
.=o.=.=Ll
15
10
LVF
5
o
Figure 3.7 (by
RVF
VISUAL FIELD
LVF
^-.
^...-
SEMANTIC
Pro\OLDGICAL
Figure 3.7 (c)
(%)
25
20
15
10
5
o
I"
PC
.=oPCPC"
I. .<
Exp7: CHARACTER
STANDARD
(%)
25
20
15
to
5
o
1.1^to SEM
CHARAorEnTYPE
Exp7:TASKvs. CHARACTER
LLl
.=
.=oP=.=LLl
,-<
(%)
25
20
15
to
5
ounl^tATED
Exp7: CHARACTERvs. VISUAL FIELD
IU
.=
CFo.=.=I"
<
Figure 3.8 (a)
STANDARD
.^..-
^.
Ex^ro SEM
CHARAorERTYPE
SEMANTIC
PHONO. coreAL
I
I',
-----
STANDARD
--n-^
--.-.
(%)
25
20
15
to
5
o
\
UNRELATED
\
1.1^ro SEM
CHARAerERTYPE
\
I^IF
LVF
Exp7: TASKvs. vFvs. CHARACTER
1.1
.=
PCo.=.=LLl
<
Figure 3.8 (by
UNRELATED
.,
,,
,
\
,,
,I
\
Figure 3.8 (c)
I
\
STANDARD
~ ./
~~,
\
Q.\I' \
I \
^--
--.-.
II
I
Fit^ro SEM
CHARAcrERTYPE
^
--.-.
SEMRVF
SEMLVF
pHORVF
pHOLVF
\
\\\
\\
UNRB. ATED Figure 3.8 (d)
}-,
F-.
o
(msec)
80
Exp6:VISUAL FIELDADVANTAGE
(CENTERED PICTURE)
>.<
~L
>. 20
60
..
LL>.=^
I- >.= .
<L
40
o
-20
>
-40
-60
-80
#I
I17
(msec)
80
#2 #3
SESSION
Exp7: VISUAL FIELD ADVANTAGE
(LATERALIZED PICTURE)
>.<
~>>
60
-I.
,L>^
40
20
. SEMANTIC
. PI. IonOLOGICAL
^,
>DC
o
#4
-20
L-40
-60
Figure 3.9 (a)
-80
#I #2 #3
SESSION
. SEMANTIC
. Pro^0010AL
#4 Figure 3.9 (by
(msec)
80
>Q<
^
LL>
>..
.
L, .>n= >^.I- <.=
Exp6: VISUAL FIELD ADVANTAGE
(CENTERED PICTURE)
60
40
20
o
-20
u. .>.= -60
-40
-80
118
#I
(msec)
80
>.<
L>.,
Exp7: VISUAL FIELD ADVANTAGE
(LATERALIZED PICTURE)
^
SESSION
LL>.
L
>
60
.
40
. SEMANTIC
. PFCNorOGICAL
.= > -20^Q
#2,3.4
20
I- <P= ,L
>P=
o
-40
-60
Figure 3.9'(a)
-80
#I
SESSION
. SEMANTIC
. PI-loneLOGicAL
#2.3,4 Figure 3.9'(b)
(msec)1100
LLl=I-Zo
<LLl.=
^
Exp 6: SESSION I TASK
1000
^
C. >
900
800
700
(msec)
1100
TASK
Exp 6: SESSION I VISUAL FIELD
SEMANTIC
LLl=F.Zo
^
1000
^
C. ><LlP=
900
Pro^coreAL
119
800
700
Figure 3.10 (a)
(msec)
1100
RVF
IU=I-
ZoI-C. ><LUP=
Exp 6: SESSION I TASKvs. VISUAL FIELD
^
1000
VISUAL FIELD
^
900
LVF
800
700
Figure 3.10 (by
RVF
VISUAL FIELD
-^
^
LVF
SEMANTIC
Pro\CLOGICAL
Figure 3.10 (c)
(msec)
1100
1.1==I-Zo
<LLlPC
^
Exp 6: SESSION I CHARACTER
1000
^
o
900
800
700
STANDARD
(msec)
1100
I"=I-Zo
<LUPC
Exp6: SESSION I TASK vs. CHARACTER
^
FCkro SEMANrlC NEurRAL
o. 1ARAcrER
1000
^
o
900
800
(msec)
1100
Exp 6: SESSION I
CHARACTER vs. VISUALFIELD
700
u,=I-Zo
<LUPC
Figure 3.1 I (a)
1000
STANDARD
^
o
900
-^
^-.
SEMEuro
CHARAerERTYPE
800
SEMANrlC
pHoiroLOGicAL
700
srANDARD Euro SEM UNRELATED
CHARAorERTYPE
Exp 6: SESSION I
TASKvs. VFvs. CHARACTER
UNRELATED
(msec)
1100
Ul=I-Zo
<LUrig
Figure 3.11 (by
^
1000
^--
^.^
^
o
900
RVF
LVF
800
,.-,,,
,,
,,
-^^.^
700
,,,,
.
Figure 3.11 (c)
SFANDARD
I
,,
I
II
I
II
I
I
I
I
I
II
\
~~.\\
^-.
--.-.
^
--.-.
\\
Fit^ro SEM
CHARAcrERTYPE
\
\.
SEMRVF
SEMLVF
PHONRVF
PHONLVF
UNRELATED Figure 3.11 (d)
I-.
I\>o
(msec)
2100
LU
=I-.Zo
<LLl.=
^
Exp7: SESSION I TASK
2000
^
<.>
1900
1800
1700
(msec)
2100
TASK
Exp7: SESSION I VISUAL FIELD
SEMANTIC
LLl
=I. .Zo
^
2000
^
,.><LLln=
1900
Pro^001^L
121
1800
1700
Figure 3.12 (a)
(msec) Exp 7: SESSION I TASKvs. VISUAL FIELD2100
RVF
U,=I-
=oI-o<1.1n=
^
2000
VISUAL FIELD
^
1900
LVF
1800
1700
Figure 3.12 (by
RVF
VISUAL FIELD
^.^
^
LVF
SEMANTIC
ProilOLOGICAL
Figure 3.12 (c)
(msec)
2200
LU=I-Zo
2100
^
Exp7: SESSION I CHARACTER
2000
^
o
1900
1800
<LLlPC
1700
1600
1500
STANDARDrK^rid^ SEMANTIC UNRELATEDFigure 3.13 (a)CHARAcrERTYPE
(msec) Exp 7: SESSION I TASKvs. CHARACTER2200
LLl=I. .=o
2100
^
2000
^
I. .o
1900
<IUP=
1800
1700
(msec)
2200
1600
1500
Exp 7: SESSION I
CHARACTER vs VISUAL FIELD
LU=I.Zo
2100
^
STANDARD
2000
^
o
1900
<LU.=
1800
^--
^
SEMrichro
CHARACTERTYPE
1700
1600
SEMANTIC
PHONOLOGICAL
1500
SFANDARD 1.10rro SEM UNRELATED
CHARAcrERTYPE
Exp 7: SESSION I
TASKvsVFvsCHARACTER
UNRELATED
(msec)
2200
LU=I-
ZoF.o<IU.=
2100
Figure 3.13 (by
2000
^
^.^
1900
1800
RVF
LVF
1700
1600
II
II
,I
II
-..-, ~
1500
IC^.- SEMRVF
--.-. SEMLVF
^- PHONRVF
--.-- PHONLVF
Figure 3.13 (c)
,IIIIIII
STANDARD
~~.
Lionro SEM
CHARAerERTYPE
^
UNRELATED Figure 3.13 (d)
-.
IQI\>
(msec)
950
LLl=I. .
=oI. ..o<IU.=
^ 900
^
,
Exp8: TASK
850
800
(msec)
950
SEMANTIC
LLl:=I-
ZoI-C. ><1.1PC
^
Exp 8: VISUAL FIELD
900
^
TASK
123
PI-^LOGICAL
850
800
(msec)
950
Figure 3.14 (a)
VISUAL FIELD
Exp 8: TASK vs. VISUAL FIELD
LU=:r-
ZoI-C. ,<LUn=
RVF
^ 900
^
850
LVF
800
Figure 3.14 (by
RVF
VISUAL FIELD
^
^
LVF
SEMANTIC
PHONOLOGIC
Figure 3.14 (c)
(msec)
1000
Ul=I. .
ZoI-o<1.1PC
950
^
Exp8: CHARACTER
900
850
800
750
700
srANDARD
(msec)
1000
1.1C^ SEM
CHARACTERTYPE
Ul
=I-
Zor-o<LLlrig
950
Exp8:TASK vs. CHARACTER
900
850
(msec) Exp 8: CHARACTER vs. VISUAL FIELD
UNRELATED
800
1000
U=I. .
ZoF.o<IU.=
750
^
950
Figure 3.15 (a)
700
^
900
STANDARD
850
800
--^..-
^.^
750
Ex^to SEM
CHARACTERTYPE
700
,..~~JP ~,
SEMANric
,III
PHONOLOGICAL
STANDARD
(msec) Exp8: TASK vs. VFvs. CHARACTER
UNRELATED
1050
Euro SEM
CHARAGrERTYPE
U,
=I. .
=oF.o<in.=
1000
^
Figure 3.15 (by
950
^ 900
^
--.-.
850
UNREIATED
Rl/F
LVF
800
I
,I _--..~,I. - ~~.I
I
750
Figure 3.15 (c)
I II I
I
, I
I
,0~,I ~~
I
STANDARD Folio SEM
CHARACTERTYPE
^--
--.-.
^--
--.-.
SEMRVF
SEM LVF
PHONRVF
PHONLVF
UNRELATED Figure 3.15 (d)
I-,
I\>J=~
(^
20
11.1I-<.=
PCo.=.=,Ll
15
Exp8:TASK
10
5
o
(^^)
20
SEMANTIC
TASK
Exp 8: VISUAL FIELD
1.1I-<PC
.=o.=.=L"
15
.
125
10
ProneLOGl^L
5
o
Figure 3.16 (a)
(^
20
RVF
LLlI-<.=
n=o.=.=IU
Exp8: TASK vs. VISUAL FIELD
15
VISUAL FIELD
10
LVF
5
o
Figure 3.16 (by
RVF
VISUAL FIELD
LVF
^.^
^.
SEMANTIC
Pro^CGICAL
Figure 3.16 (c)
(96)
20
I"I.<rig
^o.=.=LU
Exp8: CHARACTER
15
10
5
o
orANDARD
(%)
20
SEM1.1^ro
CHARAGrEn
IUr-<.=
.=o.=PCLLl
Exp8:TASK vs. CHARACTER
15
10
UNREIATED
(%)
20
5
Exp8: CHARACTER vs. VISUAL FIELD
LLlI-<.=
PCo.=.=LU
15
o
Figure 3.17 (a)
STANDARD
10
^-.-
-^.
5
1.10rro SEM
CHARACTERTYPE
SEMANTIC
pHohroLOGicAL
JP
o
,
I
STANDARD
(%)
20
UNREIATED
-^.^
--.-.
\
1.1^ro SEM
CHARAcrERTYPE
Exp8: TASK vs. vF vs. CHARACTER
IUI-
INF
LVF
.=
ECoEC.=IU
Figure 3.17 (by
15
\\\
10
\
b
UNRELATED
5
Figure 3.17 to)
.
, . \
, \, \
.~
o
STANDARD
,
I,
,
I
^.
--.-.
^0^
--.-.
ForO SEM
CHARACTERTYPE
SEMRVF
SEMLVF
PHONRVF
PHONLVF
\\ \\\\\
UNREIATED Figure 3.17 (d)
I-.,
I\>o
(msec)
80
^
L>>.
IL>n=
Exp3:VISUAL FIELD ADVANTAGE
(ISOLATING TASKS)
60
40<LL
> 20. 20
o
.
^,
I-n=
> -20.<L -40>DC
-60
-80
#I
127
(msec)
1600
#3#2
SESSION
LLl=I-.
ZoI-<,,><LLlP=
^
1550
SESSION I OF Exp. 7 & 8:
TASK vs. VISUAL FIELD
1500
^
. SEMANTIC
. Prol^:>LOGICAL
1450
#4
1400
1350
Figure 3.18
1300
RVF
VISUAL FIELD
^,^
^.-
LVF
SEMANTIC
P}IdroLOGICAL
Figure 3.19
Benton, A. L. , Hanny, H. I. , and Varney, N. R. (1975) Visual perception of line
direction in patients with unilateral brain disease. ^!:^!gg:!, 25, 907-910.
Besrier, D. , Hildebrandt, N. (1987) Orthographic and phonological codes in the oral
reading of Japanese Kana. Journal of Ex erimental Ps cholo
Bryden, M. P. and Allard, F. (1976) Visual hemified differences depend on typeface.
Memor
References
Bryden, M. P. (1977) Measuring handedness with questionnaires. ^!:^!^^, 15,
Brain and Lan ua e, ,a. , 191-200.
and Co nition, I. ^.(2), 335-343.
Bussing, A. , Bruckmann, R. & Hartje, W. (1987) Influence of figural complexity on the
617-624.
identification of Kanji and Kana characters, Cortex, 23(2), 325-330.
Cohen, G. (1972) Hemispheric differences in a letter classification task. Perce tion and
Coltheart, M. (1980) Deep dyslexia: a right-hemisphere hypothesis. In M. Coltheart, K.
Patterson & I. Marshall(Eds. ), ^^p_^^!a, . pp. 326-380. London: Routledge &
Kegan Press.
', LL(2), 139-142.
128
Colheart, M. (1983) The right hemisphere and disorders of reading. In A. Young (Ed. ),
Day, I. (1977) Right-hemisphere language processing in normal right-handers. Journal
Function of the Ri ht Cerebral Hemis here. London: Academic Press.
Learnin
Ellis, H. D. and Shepherd, I. W. (1974) Recognition of abstract and concrete words
of Ex erimental Ps cholo
103(5), 1035-1036.
Elman, J. L. , Takahashi, K. , & Tohsaku, Y. (1981) Lateral asymmetries for the
identification of concrete and abstract Kanji. ^!:^!^, 19(3), 407-4/2.
Gazzaniga, M. S. (1967) The split brain in inari. Scientific American 217 2 24-29.
presented in left and right visual fields. Journal of Ex erimental Ps cholo
Human Perce tion and Performance, ,^., 518-528.
,
Gazzani a M. S. 1970 The Bisected Brain. New York: A leton-Centur -Crofts.
Goodglass, H. and Barton, M. (1963) Handedness and differential perception of verbal
stimuli in left and right visual fields. ^^p^^s. , 17, 851-854.
Goodglass, H. and Butters, N. M. (1988) Psychobiology of cognitive processes. In R. C.
Atkmson, R. I. Hermstein, G. Lindzey, and R. D. Luce (Eds. ), Stevens' Handbook
Goryo, K. (1987) Yomu to Iukoto (Reading in Japanese). Tokyo: Tokyo University
of Ex erimental Ps cholo
Hartje, W. , Hannen, P. & Willmes, K. (1986) Effect of visual complexity in
tachistoscopic recognition of Kanji and Kana symbols by German subjects.
Press.
Hatta, T. (1976) Asynchrony of lateral onset as a factor in difference in visual field.
'a^, 2^(2). 297-300.
. (2nd ed. ) pp. 863-962. New York: John Wiley & Son.
Hatta, T. (1977) Recognition of Japanese Kanji in the left and right visual fields.
Perce tual and Motor Skills, 42, 163-166.
Hatta, T. (1977) Lateral recognition of abstract and concrete Kanji in Japanese.
'a, ,15,685-688.
Hatta, T. (1978) Recognition of Japanese Kanji and Hirakana in the left and right
visual field. Ia anese Ps cholo ical Research, 20(2), 51-59.
Hatta, T. (1979) Hemisphere asymmetries for physical and semantic congruency
129
^, 45,731-734.
^!^x, ^^Q(5), 273-278.
Hatta, T. (1981a) Different stages of Kanji processing and their relations to functional
hemisphere asymmetries. Ja anese Ps cholo ical Research, 23(I), 27-36.
Hatta, T. (1981b) Differential processing of Kanji and Kana stimuli in Japanese
matching of visually presented Kanjistimuli. The Ia anese Journal of
people: some implications from Stroop test results. ^!:^!^, 19, 87-93.
Hatta, T. (1981c) Task differences in the tachistoscopic Kanji recognition and their
relations to hemisphere asymmetries. The Ia anese Journal of Ps cholo , ,^2. (3),
Hatta, T. , Honjoh, Y. & Mito, H. (1983) Event-related potentials and reaction times as
measures of hemispheric differences for physical and semantic Kanji matching.
Cortex, 19, 517-528.
139-144.
Hayashi, R. & Hatta, T. (1982) Visual field differences in a deeper semantic
processing task with Kanji stimuli. Ja anese Ps cholo ical Research, 24(3), 1/1-
Hellige, J. B. (1976) Changes in same-different laterality patterns as a function of
practiceand stimulus quality. ^^p^^g. , 20(4), 267-273.
1/7.
Hines, D. (1976) Recognition of verbs, abstract nouns and concrete nouns from the left
and right visual half-fields. ^:^a, , 14, 211-216.
Hines, D. (1977) Differences in tachistoscopic recognition between abstract and
concrete words as a function of visual half-field and frequency. Cortex, 13, 66-
Hink, R. F. , Kaga, K. & Suzuki, I. (1980) An evoked potential correlate of reading
ideographic and phonetic Japanese scripts. ^!:^!^, 18, 455-464.
73.
Hirata, K. & Osaka, R. (1967) Tachistoscopic recognition of Japanese letter materials
in left and right visual fields. ^5.1^a, , 10, 7-18.
Hirose, T. (1984) The effect of script frequency on semantic processing of Kanji and
130
Hirose, T. (1985) The effects of orthographic familiarity on word recognition. The
Kana words. The Ia anese Journal of Ps cholo
Imura, T. (1943) Aphasia: characteristic symptoms in Japanese. Ps chiatrica at
Ia anese Journal of Ps cholo
^!:^!^a, , 47, 196-218. (in Japanese) Cited in S. Sasanuma (1980),
Acquired dyslexia in Japanese. In M. Coltheart, K. Patterson & J. Marshall
(Eds. ) ^^p_^y^!^a, pp. 48-49. London: Routledge & Kegan Press.
Inoue, M. Salto, H. & Nomura, Y. (1979) Psychological research on characteristics of
Kanji: the effects of graphemic and phonetic processing on information
, 56, 44-47.
, ^a. , 173- 176.
extraction from Kanji. ^!^!^^!!., 22, (2), 143-159. (in Japanese)
Iwata, M. (1982) Association area and Language function: especially written language.
Nihon IshikaiZasshi 88 I 63-75. in Ia anese
Iwata, M. (1984) Kanji versus Kana: Neuropsychological correlates of the Japanese
Iwata, M. , Sugishita, M. & Toyokura, Y. (1982) The Japanese writing system and
functional hemispheric specialization. In Katsuki, S. , Tsubaki, T. & Toyokura,
Y. (Eds. ), Amsterdam: Excepta Medica. pp. 53-62.
writing system. Trends in Neuro Sciences, Z(8), 290-293.
lones, E. & Aoki, C. (1988) The processing of Japanese Kanji and Kana characters, In
pp. 301-320. Berlin: Springer-Verlag.
Kawamura, M. , Hirayama, K. & Yamamoto, H. (1989) Difference interhemispheric
transfer of Kanji and Kana writing evidenced by a case with left unilateral
agraphia without apraxia. Brain, 1/2, 10/1-10/8.
de Kerckhove, D. and Lumsden, C. L. (Eds. ), The A1 habet and the Brain.
Kawamura, M. , Hirayama, K. , Hasegawa, K. , Takahashi, N. , Yamamaura, A. (1987)
Alexia with agraphia of Kanji(Japanese inorphograms). Journal of Neurolo
Kimura, D. (1961) Cerebral dominance and the perception of verbal stimuli. Canadian
13 I
^^:, 50,1125-1129.
Kimura, D. (1966) Dual functional asymmetry of the brain in visual perception.
Journal of Ps cholo
Kimura, D. (1969) Spatial localization in left and right visual fields. Canadian Journal
'a^,^., 275-285.
^^y, ^^(6), 445-458.
Kimura, Y. (1984) Concurrent vocal interference: its effects on Kana and Kanji.
, I. ^.(3), 166-171.
'a, ,33,193-201.
uarterl Journal of Ex erimental Ps cholo
Kmsbourne, M. (1973) The control of attention by interaction between the cerebral
, 36A, 1/7-131.
Kirisbourne, M. (1970) The cerebral basis of lateral asymmetries in attention. Acta
hemispheres. In S. Kornblum Ed :Attention and Performance IV. .239-256.
Kirk, R. Ex erimental Desi in Procedures for the Behavioral Sciences. (2nd ed. )
New York: Academic Press.
Belmont, Ca: Brooks/Cole.
Kitao, N. , Hatta, T. , Ishida, M. , Babazono, Y. , & Kondo, Y. (1977) concreteness,
hieroglyphicity and familiarity of Kanji. The Ia anese Journal of Ps cholo
Kittler, P. , Turkewitz, G. , and Goldberg, E. (1989) Shifts in hemispheric advantage
during familiarization with complex visual patterns. Cortex, 25, 27-32.
48, 105-1/1-
Klein, D. , MOScovitch, M. , and Vigna, C. (1976) Attentional mechanisms and perceptual
asymmetries in tachistoscopic recognition of words and faces. ^!:^!^,
Kosslyn, S. M. , Koenig, 0. , Barrett, A. , and Cave, C. B. (1989) 'Evidence for two types
of spatial representations: hemispheric specialization for categorial and
14, 55-66.
coordinate relations. Journal of Ex erimental Ps cholo
Kotani, S. (1935) A case of alexia with agraphia. Ia anese Journal of Ex erimental
Performance 154 723-735.
^3,911Q. !Q. gy, 2. , 333-348. (in Japanese) Cited in S. Sasanuma (1980) Acquired
dyslexia in Japanese. In M. Coltheart, K. Patterson, I. Marshall(Eds. ) IQ^
^!^!^a, . PP. 48-90. London: Routledge & Kegan Paul.
Mishkin, M. and Forgays, D. G. (1952) Word recognition as a function of retinal locus.
132
Mochizuki, H. , Ohmoto, R. (1988) Pure alexia in Japanese and agraphia without alexia
in Kanji. ^^^^y, 25,1157-1159.
Journal of Ex erimental Ps cholo
,
Monkawa, Y. (1981) Stroop phenomena in the Japanese language: the case of
ideographic characters (Kanji) and syllabic characters (Kana). Perce tual and
MOScovitch, M (1986) Afferent and efferent models of visual perceptual asymmetries:
Human Perce tion and
Motor Skills 53 67-77.
, 43, 43-48.
theoretical and empirical implications. ^!^a, , 24 (1), 91-1/4.
Nebes, R. D. (1974) Hemispheric specialization in coinmissurotomized inari.
Niki, R. & Ueda, S, (1977) a case of 'pure' alexia showing a remarkable recovery more
than a year after the onset. Yo o i Hos ital Medical Re ort, 14, 4-16. (in
Japanese) Cited in S. Sasanuma, Acquired dyslexia in Japanese. In M. Coltheart,
Psychological Bulletin. ^. L(I), I-14.
K. Patterson, I. Marshall(Eds. ), D^IL^a, . London: Routledge & Kegan
Ohnishi, H. & Hatta, T. (1980) Lateral differences in tachistoscopic recognition of
Paul.
Kanji-pairs with mixed image values. ^^yg^a, , 23, 233-239.
Oldfield, R. C. (1971) The assessment and analysis of handedness: the Edinburgh
inventory. ^!^^, 2, 97-1/3.
Paradis, M. , Hagiwara, H. , & Hildebrandt, N. (1985) Neuronn uistic As ects of the
Patterson, K. and Besrier, D. (1984) Is the right hemisphere literate?. ^. gg. ^:^.,^
Ia anese Writin S stem. Orlando: Academic Press.
Patterson, K. E. (1981) Neuropsychological approaches to the study of reading. British
Neuro s cholo
133
Pring, T. R. (1981) The effect of stimulus size and exposure duration on visual field
asymmetries. Cortex, 17, 227-240.
Saffran, E. M. and Mann, 0. S. M. (1977) Reading without phonology: evidence from
Journal of Ps cholo
, I. (4), 315-341.
Saito, H. (1981) Use of graphemic and phonemic encoding in reading Kanji and Kana.
aphasia.
, 72, 151-174.
Sakamoto, S. (1940) A contribution to Kanji and Kana problems in aphasia. Medical
The Ia anese Journal of Ps cholo
uarterl Journal of Ex erimental Ps cholo
Journal of Osaka Nisseki, ^., 185-212. (in Japanese) Cited in S. Sasanuma (1980),
Acquired dyslexia in Japanese. In M. Coltheart, K. Patterson, I. Marshall(Eds. ),
^12_^a, . London: Routledge & Kegan Paul.
,,^. 2. (5), 266-273. (in Japanese)
, 29, 515-525.
Sasanuma, S. & Fujimura, 0. (1971) Selective impairment of phonetic and non-
phonetic transcription of words in Japanese aphasic patient: Kana vs. Kanjiin
visual recognition and writing. Cortex, Z, I-18.
Sasanuma, S. (1974) Kanji versus Kana processing in Alexia with transient agraphia: a
case report. Cortex, 10, 89-97.
Sasanuma, S. (1975) Kana and Kanji processing in Japanese aphasics. Brain and
Sasanuma, S. & Monoi, H. (1975) The syndrome of Gigi(word-meaning) aphasia:
selective impairment of Kanji processing. ^!^!ggy, 25, 627-632.
Sasanuma, S. (1980) Acquired Dyslexia in Japanese: clinical features and underlying
mechanisms. In M. Coltheart, K. Patterson and I. Marshall(Eds. ), ^p_^^!a, .
pp. 48-90. London: Routledge & Kegan Paul.
', 2^, 369-383.
'i:+, k, F1. , Ice ",, A^., XI all?P-,',/c"..Sasanuma, SI^1980) The ature of the task-stimulus interaction in the tachistoscopic
recognition of Kanaand Kanji words. ^^^^, 2, 298-306.
Sasanuma, S. (1986) Universal and Language-specific symtomatology and treatment of
Sasanuma, S. , Itoh, M. , Mori, K. , & Kobayashi, Y. (1977) Tachistoscopic recognition of
Kana and Kanji words. ^:^a, ,15, 547-553.
Sergent, I. and Hellige, B. (1986) Role of input factors in visual-field asymmetries.
aphasia. EQ^!, 38,121-175.
134
Sergent, I. (1982) Theoretical and methodological consequences of variations in
exposure duration in visual laterality studies. ^^^g^s. , 31(5),
^^.,^., 174-199.
Shin Kokugo Jiten (1981) (New Japanese Dictionary) T. Yamada, & Y. Yoshikawa
(Eds. ), Tokyo: Kadokawa Shoten.
Sperry, R. W. (1982) Some effects of disconnecting the cerebral hemispheres. Science,
217, 1223-1226.
451-461.
Srebro, R. (1985a) Localization of cortical activity associated with visual recognition in
Srebro, R. (1985b) Localization of visually evoked cortical activity in humans. Journal
^E^^y, 360, 233-246.
humans. IQ^^, 360,247-259.
Sugishita, M. & Yoshioka, M. (1987) Visual processes in a hernialexic patient with
posterior callosalsection. ^!:^!^, 25 (2) 329-339.
Sugishita, M. , Iwata, M. , Toyokura, Y. , Yoshioka, M. & Yamada, R. (1978) Reading of
ideograms and phonograms in Japanese patients after partial coinmissurotomy.
'a, ,16,417-426.
Sugishita, M. , Yoshioka, M. & Kawamura, M. (1986) Recovery from hernialexia. Brain
^, 29,106-118.
Tanaka, Y. , Yamadori, A. & Murata, S. (1987) Selective Kana agraphia: a case report.
Cortex, 23, 679-684.
Tsunoda, T. (1985) :L!^^!!.. Tokyo: The Taishukan Publishing Co.
Tzeng, 0. L. , Hung, D. L. , Cotton, B. & Warig W. S-Y (1979) Visual lateralization effect
Waters, G. S. and Seidenberg, M. S. (1985) Spelling-sound effects in reading: Time-course
and decision criteria. ^^^!!., 13, 557-572.
Yamadori, A. (1975) Ideogram reading in alexia. Brain, 98, 231-238.
in reading Chinese characters' Nature, 282, 499-501.
135
Yokosawa, K. & Umeda, M. (1988) Processing in human Kanji-word recognition.
Proceedin of the 1988 IEEE International Conference on S stems Man and
Zaidel, E. (1978) Lexical organization in the right hemisphere. In P. Buser & A. Rougeul-
Buser (Eds) :Cerebral Correlates of Conscious Ex erience. Amsterdam: Elsevier.
"^, 377-480.
Zaidel, E. & Peters. A. (1981) Phonological encoding and ideographic reading by the
disconnected right hemisphere: Two case studies. ^z^!^^, 14, 205-
Zusetsu Nihong0 (1982) Miyajima, T. Nomura, M. , Egawa, K. , Nakano, Y. , Sariada, N. ,
Satake, H. (Eds. ) Kadokawa Sho Jiten v01 9. Tokyo: Kadokawa Shoten.
234.
Name: Chisato Aoki
Date and Place of Birth: November 23, 1956; Shizuoka, Japan
BIOGRAPHICAL DATA
College: Doshisha University, B. A. , English literature, 1979
Graduate Work: Northeastern University, M. Ed. , Educational Research, 1983
Northeastern University, M. A. , Psychology, 1985
136
Top Related