I I I I I {LII-I.

'Hfimg

SITY LIBRAHIE

IlllUllllHUIlllllllllllllllllllllll'lull lllllslllll

3 1293 00896

 

 

 

 

 

 

This is to certify that the

dissertation entitled

AN ASSESSMENT OF DIEFICULTY IN JAPANESE TEXT
IN RELATION TO COMPUTER-MEDIATED TEXT

presented by

William Server

has been accepted towards fulfillment
of the requirements for

 

Ph.D. degree in Linguistics

 

Major professor

Date December 12, 1991

 

M5 U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

___.,.—_f

4 .. ____

 

u..—

7

 

IL

LIBRARY

University

Michigan State

F‘

r
l

 

.1

*

TO AVOID FINES return on or before one due.

I PLACE IN RETURN BOX to remove this checkout from your record.

 

 

DATE DUE DATE DUE DATE DUE

 

 

3m 1 o 1996

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity lnetltutlon

chS—M

 

An Assessment of Difficulty in Japanese Text
in Relation to Computer-Mediated Text

by

William Carl Server

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Linguistics

1991

.ABSTTUKIP

AN ASSESSMENT OF DIFFICULTY IN JAPANESE TEXT
IN RELATION TO COMPUTER-MEDIATED TEXT

By

William Carl Server

The main purpose of this pilot study was to examine the
effect of syntactic variables on the reading of Japanese as a
second language. This study addressed the following two
hypotheses: 1) grammatical variables affect the difficulty
of the reading of a Japanese text by native speakers of
English; and 2) computer-mediated text can facilitate the
comprehension of a passage more than ordinary text.

In considering the first hypothesis, this pilot study
focused on the effects of grammatical structure on passage
difficulty. As a part of the study, the contributions of
rhetorical structure, vocabulary, and culture to textual
difficulty were also considered in order to enlarge the total
picture of the major factors that affect the level of
difficulty of Japanese text. All these factors help
contribute to the difficulty that faces the reader of
Japanese as a second language. This pilot study considered a
list of 20 categories by which passage difficulty can be

measured. This list is an enumeration of 20 possible

variables that are hypothesized to influence passage
difficulty. In order to explore the second hypothesis,
computer software was designed with the 20 variables in mind
in order to determine whether the software can be used to 1)
measure quantitatively the contribution these factors make to
passage difficulty and 2) actually facilitate the reading of
Japanese text. This software was tested on human subjects in
order to measure its effectiveness in measuring textual
difficulty and its power to facilitate the comprehension of
Japanese written text.

Based upon the results of this study, syntactic
variables used in the research, such as tense, subordination,
passive, causative, quotations, and relative clauses, were
shown to affect the difficulty involved in reading Japanese
text by native speakers of English, thus confirming the first
hypothesis. However, computer-mediated text was not shown to
facilitate the comprehension of Japanese text more than
ordinary text, thus leaving the second hypothesis

unconfirmed.

ACKNOWLEDGMENTS

I would like to express my gratitude to Professor Ruth
Brend for guiding the analysis in this dissertation. I am
also indebted to Professor John Eulenberg, Professor Paul
Munsell, Professor Mutsuko Simon, and Professor Seok Song for
their help and encouragement. I am also grateful to
Professor Hiroshi Nara of the University of Pittsburgh for
his assistance in finding a topic. I would like to thank Mr.
Yasuo watanabe for allowing me access to his students for the
empirical testing. Finally, I would like to acknowledge the
love and support of my wife, Hea Young, without whom this

research would have been impossible.

TABLE OF CONTENTS

List of Charts.

List of Tables.

List of Figures.

1.

The Contribution of the Computer to this Pilot

The Gap in Japanese L2 Reading Research.

The Evaluation of Difficulty in Two Texts.
An Experiment Assessing Difficulty in the Two

Features Counted in the Estimation of Reading

The Particle
The Particle
The Particle
The Particle

hi.
to.
o.

no.

Introduction.
1.0. Purpose.
1.1. Grammatical Considerations.
1.2. Other Considerations.
1.3.
Study.
1.4. Summary.
Review of Literature.
2.0. Reading Theory.
2.1. Contrastive Rhetoric.
2.2. Readability Formulas
2.3. Culture and Reading Comprehension.
2.4. Computers and Computer-mediated Text.
2.5. Japanese Textual Variables.
2.6.
Method.
3.0. Purpose.
3.1. Variables in this Study.
3.2.
3.3.
Texts.
3.4. Computer-mediated text.
3.5. A Test of the Software.
3.6. Summary.
Assessment of Difficulty in Japanese Text.
4.0. Introduction.
4.1.
Difficulty.
4.1.1. Homophonous Particles.

4.1.1.1.

4.1.1.2.

4.1.1.3.

4.1.1.4.

4.1.1.5.

The Particle

4.1.2. Nominalizations.

4.1.3. Anaphora.

4.1.4. Tense.

mo

viii

ix

xi

UNI-'H

4.4.

The Experiment to Assess Difficulty in Japanese Text.

5.0.
5.1.

5.2.

The Computer Experiment.

6.0.
6.1.

6.2.

6.3.

O‘U'lIFMNI—IOIOeee
eeeeeee

HHHI—‘HHHHHHHHI—‘H
q
0

Hi—Ii—IHI—H—H—H—IHomqmm

a)
e

thhhfibbhkfihhbkfi

H
N
o
0

4.3.1.
4.3.2.

Summary.

WOrds per Sentence.
Noun Head Modification.
Conjunctions.
Subordinators.

The Passive.

Causative.

Verb Compounding.

Quotations.

Relative Clauses.

Hedges.

Further Embeddings.

Comparatives.

Rhetorical Structure.

The Difficult Kanji in a Passage.

Vocabulary.

Cultural Assumptions.

The Evaluation of Two Texts.

The Results of the Evaluation.

The Results for "Aruke Orinpikku".

The Results for "HOogen to Kyootsuugo".

Introduction.

Data Presentation.
The Analysis of
The Analysis of

5.1.1.
5.1.2.

Conclusions.

5.2.1.
5.2.2.

Conclusions for
Conclusions for

Introduction.
The Program.
The Introduction to the Program for

"Aruke Orinpikku"

"Aruke Orinpikku".

The Grammar Section.
The Rhetoric Section.
The Culture Section.

On-line Help Received.

Passage Reading Time.

6.1.1.

the User.
6.1.2. The Vocabulary.
6.1.3. The Kanji Section.
6.1.4.
6.1.5.
6.1.6.
6.1.7. The Tests.
6.1.8. Monitoring.
6.1.9. The Procedure.
The Results.
6.2.1.
6.2.2. Test Scores.
6.2.3.
Discussion.

vi

"Hoogen to KYootsuugo".

"HOogen to Kyootsuugo".

168
177

189
189
190

191
195
198
200
202
204
206
208
209
211
212
212
213
213

7. Discussion.
7.0. Introduction.
7.1. The Study.
7.2. The Results.
7.2.1. The Implications of the Data of

Chapter 5.

7.2.1.1. The Implications for "Aruke
Orinpikku".

7.2.1.2. The Implications for "Ebogen
to Kyootsuugo".

7.2.1.3. Summary of the Implications.

7.2.1.4. Conclusions.

7.2.2. The Implications of the Data of

Chapter 6.

7.2.2.1. The Subjects.

7.2.2.2. The Contributions to Passage
Difficulty.

7.2.2.3. Discussion of the Test
Scores.

7.3. Conclusion.

Appendix A.

Appendix B.

Appendix C.

Bibliography.

The Passages Analyzed According to the
Criteria in Chapter 4.

”Aruke Orinpikku".

"Hoogen to Kyootsuugo"

The Passages Written in Japanese
Orthography.

”Aruke Orinpikku".

”Reagan to Kyootsuugo".

The Comprehension Questions.

"Aruke Orinpikku".
"Hoogen to Kyootsuugo”.

vii

215
215
216
220

220
220
227
235
237

238
238

239

242
243

245
245
256

261
261
262

263
263
269

275

Chart

Chart

Chart
Chart

Chart

Chart
Chart

Chart

2:

3:
4:
5:

7:
8:

LIST OF CHARTS

Percentages of Predicted Problems in "Aruke
Orinpikku".

Percentages of Predicted Problems in "Ebogen
to Kyootsuugo".

”Aruke Orinpikku" Predictions Chart.
WAruke Orinpikku" Actual Problems Chart.

Percentage of Actual Problems in "Aruke
Orinpikku".

”Ebogen to Kyootsuugo” Predictions Chart.
”Hbogen to Kyootsuugo” Actual Problems Chart.

Percentage of Actual Problems in "HOogen to
Kyootsuugo".

V1“

97

101
171
174

176
179
183

186

Table
Table
Table
Table

Table
Table
Table
Table
Table
Table
Table
Table

Table

Table

Table

Table

Table

Table

4:

10:
11:
12:

13:

14:

15:

16:

17:

18:

LIST OF TABLES

TAruke Orinpikku" Problems.
"Reagan to Kyootsuugo" Problems.
Difficulty Indices for Both Passages.

Percentages of Predicted Errors for each
Passage.

tAruke Orinpikku" Predictions.

“Aruke Orinpikku" Problems.

”Ebogen to Kyootsuugo" Predictions.
"Hoogen to Kyootsuugo" Actual Problems.
Percentages of Help Requested.

Test Scores.

Passage Reading Time.

"Aruke Orinpikku" Predictions and Actual
Problems.

The Predicted Success Rate and Actual Success
Rate for "Aruke Orinpikku".

The Predicted Influence and Actual Influence
for FAruke Orinpikku".

"Ecogen to Kyootsuugo" Predictions and Actual
Problems.

The Predicted Success Rate and Actual Success
Rate for "Hoogen to Kyootsuugo".

The Predicted Influence and Actual Influence
for ”Boogen to Kyootsuugo".

Difficulty Indices for "Aruke Orinpikku" and
"Hoogen to Kyootsuugo".

98
102
104

105
170
173
178
182
212
213
213

220

221

223

227

228

230

236

Table 19: Percentages of Difficulty for "Aruke
Orinpikku". 239
Table 20: Percentages of Difficulty for "Reagan
to Kyootsuugo". 241

LIST OF FIGURES

Figure : The Text Page for "Ecogen to Kyootsuugo". 192

Figure 2: The Vocabulary Page for "Boogen to
Kyootsuugo". 194

Figure : Defining wari ni in "HOogen to Kyootsuugo". 196
Figure : Defining hoogen ”HOogen to Kyootsuugo". 197

Figure 5: Defining the kanji in (moyoo)shi in the
Passage, WAruke Orinpikku". 199

Figure 6: A.verb in the Passive in "Aruke Orinpikku". 201
Figure 7: Defining the ten in "Aruke Orinpikku". 203
Figure 8: A culture Explanation in "Aruke Orinpikku". 205

Figure 9: Comprehension Question 1 for "HOogen to
Kyootsuugo". 207

xi

CHAPTER 1
Introduction

1.0. Purpose. This pilot study attempts to assess
difficulty in Japanese texts in relation to reading
comprehension. Most current models of reading comprehension
take an interactive point of view. The interaction comes
from the active participation of the reader in negotiating
both top-down and bottom-up reading strategies. Top-down
reading strategies start with the knowledge the reader brings
to the reading task including a knowledge of the world, the
domain of discourse, and grammar. The reader uses this
knowledge in actively engaging the text, which is perceived
through the use of bottompup reading strategies involving the
identification of the distinctive patterns of letters,
lexical items, grammatical structure, and rhetorical
organization. These patterns are interpreted through the use
of the reader's given knowledge which he or she brings to the
task. Thus, through the use of top-down and bottom-up
reading strategies, the reader actively negotiates the
meaning of a text.

The purpose of this dissertation is to consider the
contribution of the structure of written text to the reading
process, in particular to the level of difficulty that it

poses. The level of difficulty involved in reading a

particular text is an interaction of many different factors.
One set of factors involves the implementation of top-down
reading strategies. Knowledge of the universe of discourse
is certainly one important factor in determining the level of
difficulty of a text. It is impossible to fully understand a
reading on sumo wrestling in Japanese without a cultural
understanding of this sport. Furthermore, a knowledge of
rhetorical structures which a reader brings to a reading task
facilitates the comprehension process. Knowing the typical
structure of a personal or business letter, newspaper
article, or postcard aids in their comprehension. Also, in
addition to this, without a firm grounding in syntactic
structure, the reader cannot comprehend text. As can be seen
from these examples, the knowledge the reader brings to a
reading passage is crucial in interpreting text.

Another set of factors involves the implementation of
bottom-up strategies, which relate to the identification of
textual cues. Once the textual cues are recognized through
bottom-up strategies, top-down strategies interpret these
cues to bring meaning to the text. In this process, textual
structure is interpreted in light of the a priori knowledge
the reader brings to the comprehension task.

1.1. Grammatical Considerations. This pilot study focuses
the contribution of linguistic features, such as

nominalization, anaphora, tense, embedding, and case grammar,

to overall textual difficulty. Other factors that contribute
to textual difficulty are considered in this study, such as
world knowledge, knowledge of vocabulary, or knowledge of
rhetorical structure, but the relationship between the
linguistic features mentioned above and readability is the
main consideration of this study. Linguistic structure has
not been taken into account by readability formulas, such as
the Flesch Index and the Dale-Chall Index (Rude, 1986, p.
12 9 ) - The base for the calculation of readability should be
expanded to include linguistic structure in the measurement
Of textual complexity since this structure contributes to
reading difficulty. This study is an attempt to expand this
base -
1 - 2 - Other Considerations. Discourse structure also has a
hear ing on the comprehensibility of Japanese text. The main
data for this study is expository discourse. Japanese
exPository discourse has a basic organization called the tri-
shoca-ten-ketsu where ki means 'beginning', shoo
'development', ten 'change', and ketsu 'conclusion' (Hinds,
198 3a, p. 188). The basic pattern of Japanese expository
discourse is to have a beginning, a development, a change to
a related theme, and a return to a conclusion. English
exPcsitory rhetorical structure does not always have a ten
8trheture where there is a change to a subtopic that has an

indirect relationship with the original topic. As a result,

an English-speaking reader may view Japanese expository text

as having a lack of coherence, which would help to impede
understanding. In view of these considerations, Japanese
rhetorical structure can be seen as a source of problems for
comprehension on the part of English-speaking readers of
Japanese.

Another consideration in evaluating textual difficulty
for readers of Japanese as a second language is the
dif ficulty of the Japanese writing system. Written Japanese
is a mixture of a system called kana, consisting of both
hira gana and katakana subsystems, and Chinese characters,
Cal led kanji. The more difficult the Chinese characters used
in a. text, the more difficult the text becomes. The level of
difficulty of Chinese characters is primarily measured in
terms of frequency of use. The more frequently readers are

eilil><>sed to particular Chinese characters the easier they are
to learn. The difficulty of Chinese characters is certainly
a factor in determining general textual difficulty.
According to the literature (Anderson and Freebody,
1981; Coleman, 1971; Freebody and Anderson, 1981; Klare,
1974—1975; Lado, 1972; Marks, Doctorow, and Wittrock, 1974;
Thorndike, 1973), vocabulary difficulty is a strong predictor
of overall textual difficulty. The main means of measuring
voc«Elbulary difficulty is through the estimate of word

erQuency. However, the reliability of word frequency

estimates is an important issue in the accuracy of

predictions with regard to difficult vocabulary items (see
section 2.2). In considering overall textual difficulty,
vocabulary difficulty is an important consideration, but
needs to be balanced against other considerations, such as
linguistic and rhetorical structure.

Another important factor in determining textual
difficulty is the general background of the passage, which
includes knowledge of the subject content behind the passage
and also the cultural matrix into which the passage fits.
The content of a passage is the subject matter which is being
Written about. To understand a car repair manual, it is
necessary to understand the component parts that make up an
antonobile and how they work together. In addition, each
mode]. of a car has its own idiosyncrasies. Without a
background knowledge of how cars work, it is very difficult

to understand a manual on how to repair them. A more global
<3°Ifl8:ideration is the culture which forms the matrix into
which a passage fits. The example of sumo wrestling has
311‘ eady been mentioned as an example of the importance of
cultural understanding. Barnitz (1986, p. 100-101) discussed
the research by Andersson (1981), Andersson and Gipe (1983)
and Reynolds, Taylor, Steffenson, Shirey, and Anderson (1981)
which shows that when subjects read texts related to their

Own culture, they exhibit greater recall and inferencing

power. Both the content of text and the culture expressed in
it add difficulty through the concepts conveyed by the
vocabulary.

As can be seen, there are a number factors involved in
the consideration of textual difficulty in Japanese: the
linguistic structure, the rhetorical structure, the Chinese
characters, the vocabulary, and the background of the text
including the subject matter and cultural setting. (As stated
before, the focus of this pilot study is on linguistic
difficulty, but this is not to say that other factors are
neglected. They, too, are considered. As it turns out,
vocabulary and Chinese characters are important
considerations in the evaluation of textual difficulty for L2
readers of Japanese. It is important that a focused but
balanced consideration of the factors involved in textual
difficulty be given. This is the purpose of this pilot
study.

1.3. The Contribution of the Computer to this Pilot Study.
The computer can present text on a screen which is linked to
on-line help resources through a hypertext environment. The
hypertext utilized in this study enables an L2 reader of
Japanese text to choose a word, a phrase, a sentence, or a
larger unit on the screen and have it defined, translated, or
explained. Such an environment constitutes computer-mediated

text. The resources that are available facilitate the

understanding of the text. These resources are presented on
a help menu, and the reader merely chooses the kind of help
that is needed to understand a textual element. The purpose
of computer-mediated text is to provide the most appropriate
help to the user.

The help menu presents certain choices to facilitate
comprehension. It provides assistance in the following
areas: 1) vocabulary, 2) Chinese characters, 3) grammar, 4)
rhetorical structure, and 5) culture. The grammatical
information given in the program is the most complete since
grammar was the focus of this study. Predicted areas of
textual difficulty are underlined in the presentation of
computer-mediated text, so they can be selected to provide
the exact help needed. Computer-mediated text makes
appropriate help available in order to facilitate textual
understanding.

The computer is also used as a monitor of the kind of
information the reader accesses to understand the text while
reading a passage. It can keep track of how often and how
much information on vocabulary, Chinese characters, grammar,
rhetoric, and culture is used. An analysis was made of this
information to determine what part of the text seems to cause
the most difficulty. It is assumed that the more difficult a
certain textual element is, the more often information on

that element might be expected to be accessed in order to

understand it; thus, difficulty can be measured by the number
of times information on certain textual elements is accessed.
This information can be used to understand the sources of
textual difficulty.

Basically, this pilot study uses the computer as a means
of instruction in two traditional modes: 1) the inquiry mode
and 2) the tutorial mode. In the inquiry mode, the
interaction with the computer is user-initiated. The
information stored in the computer is accessed as the user
works his or her way through the computer lesson. This
information is available on request. A menu of available
assistance is on the screen to help the student if he or she
should run into trouble while working his or her way through
a computer lesson, . All the user need do is make an enquiry
by selecting options on the screen that provide the needed
information. In this mode, the user has a degree of freedom
in the interaction with the computer in that the user can
decide whether or not to query the computer for specific
information.

In the tutorial mode, there is no such freedom. The
computer asks the questions which the user must answer or
makes the requests which the user must respond to. Then the
computer checks the answers or responses and branches

appropriately. In this mode, the user has little control

over the computer, which decides what is to be studied and
how it is to be taught.

As stated above, the program written for this pilot
study contains elements of both these modes. First of all,
the text is presented in the inquiry mode because the
passages that form the reading material in this program are
presented in a hypertext environment, in which, the user can
choose a word or larger unit on the screen and have the
program define, translate, or explain it. The information on
these words or larger units has been placed into a kind of
data base which the user can access when needed. In this
kind of environment, the interaction with the computer is
user-initiated. Because of this, the user can move through
the learning material at his or her own pace making it ideal
for presenting reading materials. The user can also choose
the order in which material is covered. If the user is
having trouble with certain words, phrases, and clauses in
the text, he or she can choose the information necessary to
receive definitions, explanations, and other forms of help at
the appropriate time. The hypertext system gives the user
power to choose the appropriate help needed to understand
textual material. Hypertext can provide help that is
convenient and easily-accessible. The user just needs to
choose the help that is necessary to understand the text.

This is the essence of the inquiry mode. Secondly, the

tutorial mode in this program is present in the on-line tests
that are given after each respective text is read by the
user. The computer presents the questions which then the
user answers. At this point, there is no opportunity to make
queries in order to gain further help or information. The
only appropriate response on the part of the user is to
answer the questions. In this case, the computer decides
what is to be studied and how the user is to respond. Both
the inquiry and the tutorial modes work together in the
program written for this pilot study to form a learning
environment where the user is both given the opportunity to
get the assistance necessary to understand Japanese text but
also is tested to see how well he or she understood the text.
This creates a learning environment that is both supportive
and challenging.

1.4. Summary. This was a pilot study involving an on-going
project to study how linguistic complexity affects the
difficulty of Japanese text when it is read by native
speakers of English. Along with linguistic complexity, the
role of vocabulary, rhetorical structure, Chinese characters,
and culture were considered as to their contribution to
passage difficulty. The role of computer-mediated text in
facilitating textual comprehension was also considered in
light of how it can help the reader understand linguistic

complexity by providing explanations of the structure of

IO

Japanese sentences. This information can help the reader
parse the sentences he or she is processing so that with the
aid of information on vocabulary, rhetorical structure,
Chinese characters, and culture, he or she can understand the
text. The next chapter is a discussion of the relevant

literature that forms a background for this pilot study.

CHAPTER 2
Review of Literature

2.0. Reading Theory. Passage difficulty is related to the
knowledge that the reader brings to the task of comprehending
a text. According to James (1987, p. 177), this knowledge is
expressed in the form of schemata. Schemata represent a
nested system of cues which enable the reader to get the
maximum amount of processing from the smallest amount of
information. These schemata incorporate grammatical,
lexical, rhetorical, or cultural information (ibid., pp. 177-
181). Linguistic schemata relate to the knowledge of the
grammar the text is written in. Lexical schemata represent
the structure of vocabulary and the shape of characters used
for writing. Content schemata reflect the knowledge about
the topic the reader brings to the reading task. Formal
schemata characterize the knowledge the reader has of
rhetorical structures. Cultural schemata represent the
knowledge the reader has about the culture expressed in the
text. Based upon the analysis of James, the difficulty of
reading a passage can be seen to vary in proportion to the
k“min-edge the reader has of its grammatical structure,
vocabulary, graphemic structure, content, rhetorical
structure, and cultural content. Furthermore, Eskey (1986,

13' 13) organizes the knowledge brought to the reading task

into higher-level and lower-level cognitive skills involving
higher-level and lower-level schemata. Upper level schemata
involve a general knowledge of the world including culture
and the subject matter of the passage. Lower level schemata
have to do with the knowledge of form necessary for textual
recognition involving grammatical features and lexical
meanings. These schemata are matched with the textual
features in order to facilitate interpretation.

Most current models of reading comprehension take an
interactive point of view (Eskey, 1986; Mitchell, 1982;
Pearson, 1984). First of all, "reading is an interactive
process between reader and text" (Carrell, 1987, p. 2).
Secondly, there is an "interaction of the reader's several
kinds of knowledge" as contained in the schematic structure
of his or her background knowledge (Eskey, 1986, p. 16).
This interaction comes from the active participation of the
reader in negotiating between top-down and bottom-up reading
strategies. With data-driven bottom-up processing, the
incoming data activate the lower level schemata having to do
with grapho-phonemic, lexical, grammatical, and rhetorical
recognition (Eskey, 1986, p. 18). According to Carrell and
Eisterhold (1983, p. 557), since "schemata are hierarchically
Organized, from most general at the top to most specific at
the bOttom” , the ”bottom-level schemata converge into higher-

leve1, more general schemata" which ”too become activated".

I3

Bottomsup strategies have to do with a knowledge of form and
the recognition of visual cues in the text. Conceptually-
driven top-down reading strategies start with the knowledge
the reader brings to the reading task including a knowledge
of the world, the domain of discourse, the culture, and
pragmatics. The reader "makes general predictions based on
higher level, general schemata and then searches the input
for information to fit into these partially satisfied, higher
order schemata” (ibid.). According to Rumelhart (1980, p.
42), this processing can occur "in both directions”, either
top-down or bottom-up. The reader uses his knowledge of the
world, the domain of discourse, the culture, and pragmatics
in actively engaging the text, which is perceived through the
use of bottom-up reading strategies involving the recognition
of the distinctive patterns of letters, vocabulary, syntax,
and rhetoric. These patterns are interpreted though the use
of the knowledge the reader brings to the task. Goodman
(1967, p. 108) describes the top-down process as a
"psycholinguistic guessing’ game" 'which "involves an
interaction between thought and language". This interaction
is described as follows:

Reading is a selective process. It involves partial

use of available minimal language cues selected from

Perceptual input on the basis of the reader' 8

expectation. As this partial information is processed,

tentative decisions are made, to be confirmed, rejected

or refined as reading progresses. More simply stated,

reading is a psycholinguistic guessing game. It
involves an interaction between thought and language.

I4

Efficient reading does not result from precise
perception and identification of all elements, but from
skill in selecting the fewest, most productive cues
necessary to produce guesses which are right the first
t1me.
According to a contemporary of Goodman, Smith (1971, 1982),
what happens behind the eye is more important than what the
eye does during the reading process. Smith supports Goodman
in asserting that the reader is actively involved in
interpreting a set of visual cues in order to get the maximum
amount of information from the text.

Through the simultaneous processing of top-down and
bottom-up reading strategies, the reader actively constructs
the meaning of a text. It is through this interaction that
information is conveyed from the text to the reader. It is
important to note that the expressions, "top-down" and
"bottom-up", refer to the reading strategies and not to the
linguistic structures themselves.

2.1. Contrastive Rhetoric. The expository style used in
Japanese rhetoric is different from the expository style in
English. Discourse constraints involving discourse style
also affect the comprehensibility of a Japanese text. The
texts discussed in this dissertation are expository in
nature. As mentioned in chapter 1, John Hinds (1983a, p.
188) has stated that the basic organization used for
rhetorical structure in Japanese is ki-shoo-ten-ketsu where
*1 means 'beginning', shoo 'support', ten 'change', and ketsu

'conclusion'. The ki begins the topic, the shoo>develops it,

IS

the ten moves off the track of the topic to a related area,
and the ketsu ties everything together in the form of a
conclusion. However, this is not a conclusion in the shape
of a summary but in terms of a global statement that brings
the argument to a new level. This style of expository
writing is highly valued by native Japanese speakers (Hinds,
1983a, p. 194).

According to Hinds (1990, p. 89), there are basically
two styles of writing in English: deductive and inductive.
Deductive writing has the thesis statement in the initial
position of a composition while inductive writing has the
thesis in the final position. In English, expository
compositions that follow these basic models are seen as
having unity, focus, and coherence; in other words, they are
considered well-organized. However, according to Hinds
(ibid, p. 98), other languages may have other forms of
expository writing that appear logical and coherent to native
Speakers of those languages but to speakers of English appear
to lack organization. The forms of expository writing in
other languages may appear to be inductive since they do not
aPPt-zar to have a clear thesis statement at the beginning of
the exposition. Hinds (ibid., p. 99) states that this is
because native speakers of English tend to think that
anything that is not deductive must be inductive since they

Operate on this basic dichotomy. From this, Hinds (ibid.,

l6

pp. 99-100) concludes that in order to understand forms of
writing in other languages, it is important for the English-
speaking reader to realize that other languages may not
operate on the basis of the deductive-inductive dichotomy.
This is because good expository writing in languages other
than English may be ordered along different lines.

Expository writing in many Asian languages, such as
Japanese, Chinese, Korean, and Thai, are organized in style
according to a different principle, according to Hinds (1990,
p. 98). This style is neither deductive (a thesis statement
followed by arguments to back it up) nor inductive (a series
of arguments leading up to a thesis statement), but rather
quasi-inductive which "has as its purpose the task of getting
readers to think for themselves, to consider the observations
made, and to draw'their own conclusions” (ibid., pp. 99-100).
The nature of this kind of writing in these Asiatic languages
he states, is reader-oriented in that the reader is supposed
to make a greater contribution in understanding a passage
than is an English-speaking reader; however, an English-
sPeaking reader who reads one of these Asiatic languages may
find it incoherent. Ebbit and Ebbit (1982, p. 393) as quoted
in Hinds (1990, p. 98) define coherence in English exposition

as fOIIOws 3

Coherence--the traditional name for relationship,

Connection, consecutiveness--is essential in expository
Writing. It's essential because you can't count on the
minds of others working the same way as your mind works.

17

You must guide your readers from one idea, from one
sentence, to another. To make a coherent presentation
you have to arrange your ideas so that others can
understand them.

According to Hinds (1990, p. 99), for the English-
speaking reader, this means that if an expository text is
deductive, he or she expects the thesis to be stated clearly
and then justified clearly point-by-point in a logical
sequence of supporting material. Hinds (ibid.) further
states that if the material is not presented in a deductive
fashion, then the English-speaking reader assumes that the
text that he or she is reading is inductive and that the
sentences arranged are arranged in a clear logical order of
reasons leading to a final conclusion. Hinds (ibid.) says
the English-speaking reader expects that the arguments will
support the conclusion in a focused and coherent fashion.
Based upon the analysis of Hinds, it can be seen that when an
English-speaking reader is confronted with more loosely
organized texts (as found in Asiatic languages) where the
reader is expected to make more of a contribution in drawing
the loose ends together, the reader will naturally be
confused. Thus, according to Hinds (ibid., p. 90), the
Quasi-inductive style in Japanese appears to lack coherence
‘30 a reader from an English-background although to the
J'ilPanese native speaker it is perfectly coherent. Hinds

(ibid., pp. 99-100) finally concludes that the quasi-

inductive approach merely assumes that the reader will take

18

more of an active role in getting comprehensible meaning out
of a passage. In fact, Hinds (1987, p. 143), makes the
following proposal about writer and reader responsibility:
I suggest a typology that is based on speaker and/or
writer responsibility as opposed to listener and/or
reader responsibility. What this means is that in some
languages, such as English, the person primarily
responsible for effective communication is the speaker,
while in other languages, such as Japanese, the person
primarily responsible for effective communication is the
listener.
Because of this typology, the English-speaking reader needs
to take what seems to the Western reader a more active role
while reading Japanese because in Japanese and in other
languages (e.g., Korean), at times, "the conclusions do not
seem to follow from the reasons that lead up to them" (Hinds,
1990, p. 99). According to Hinds (ibid.), whether an
expository composition is deductive or inductive, the
English-speaking reader expects each reason presented to
support the conclusion; however, this expectation is not met
in these Asiatic languages. Hinds (ibid., pp. 99-100)
further states that with respect to these languages, the
reader needs to recognize that a set of observations will be
Organized around a loosely defined topic and that he or she
is expected to sort out these observations, make sense of
them, and make his or her own inferences. Hinds (ibid., p.
100) Concludes by observing that with respect to these

Asiatic languages, the author's job is not to argue

Persuasive1y or to "convince", but to suggest and persuade

l9

the reader to consider the issues involved and to come to his
or her own opinion. In this point of view, the reader
assumes more responsibility in coming to grips with the topic
put forth by the text.

From these observations, it is evident that one's own
expectations and assumptions as to logic, audience, and the
responsibilities of the reader come from one's own native
language and culture. This is a problem with the identity
fostered by one's own language and culture. When an English-
speaking person reads a second language, such as Japanese, or
Korean,he or she needs to be aware of the fact that the
material was written by people with a different cultural
identity which also determines how exposition is written and
presented (Shen, 1989, pp. 459-466). Contrastive rhetoric is
concerned with the knowledge that makes up the identity of
the writer in order to help the L2 reader know what to
expect. Among the kinds of knowledge the reader needs to be
aware of are (Grabe and Kaplan, 1989, pp. 271-272):

1) ”knowledge of rhetorical patterns of arrangement and the
relative frequency of various patterns"; 2) ”knowledge of the
mOrphosyntax of the target language, particularly as it
aPPlies at the intersentential level”; 3) "knowledge of the
Coherence-creating mechanisms of the target language" (target
language = L2); 4) "knowledge of the writing conventions of

the target language in the sense of both frequency and

20

distribution of types and text appearance (e.g., letter,
essay, report, etc.)"; 5) "knowledge of audience
characteristics and expectations in the target culture"; and
6) ”knowledge of the subject to be discussed, including both
”what everybody knows" in the target culture and specialist
knowledge. When the L2 reader is aware of these kinds of
knowledge, the expectations of the reader will more closely
coincide with the intentions of the writer, thus facilitating
understanding and lowering the level of frustration.
Eggington (1987, p. 166) relates the importance of
expectations to the reading theory expressed by Goodman
(1967, p. 108) in his description of the reading process as a
psycholinguistic guessing game. Eggington states that if
reading is a psycholinguistic guessing game, ”then readers
and writers have certain built-in expectations about the
ordering of ideas in any stretch of discourse" (1987, p.
166). The closer the reader's expectations of what comes
next match those of the writer, the more the communication
process is enhanced. The purpose of contrastive rhetoric is
to tailor this knowledge to the needs of the L2 reader to aid
in his or her comprehension of the text.

2.2. Readability Formulas. This dissertation focuses on the
structural aspects of texts as these aspects make an overall
contribution to the general difficulty of a text.

Readability estimates have been created to determine whether

21

textual material is at an appropriate level for the reader.
Typically, readability formulas that have been discussed in
the literature have not taken into account linguistic
structure. A list of readability formulas is given by Rude
(1986, p. 130):

1. The Flesch Index--designed to analyze material from
fifth-grade through college graduate level. Uses a
syllable per sentence average to determine grade
elevel.

2. The Fog Index--designed to analyze grade 6 through
graduate-level reading material. It is based on the
number of words of three syllables or more as an
indicator of difficulty.

3. The Dale-Chall Index--designed to measure fourth-
grade through college materials. It is based on two
types of counts: average sentence length and
percentage of unfamiliar words (i.e., those words
that do not appear on the Dale List of 3,000 Commonly
Used WOrds).

4. The SMOG Index--an estimate of grade level that a
person must attain to comprehend the text that is
being analyzed. This analysis assumes that ten
consecutive sentences will be used in the
computation.

5. The Wheeler-Smith Index--a primary-grade analysis
that is based on the number of polysyllabic words
(i.e., three or more syllables).

6. The Spache Index--based on the number of words and
sentences and affected by words not on Stone's
revision of the Dale-Chall 769 Easy word List.

Geoffrion and Geoffrion (1983, p. 104) also characterize the
Spache and Dale-Chall formulas as combining "average sentence
length with the percentage of rare words”. They also mention
the Fry Formula, which like the Flesch Count utilizes
sentence length and the number of syllables in 100 words of
text (ibid.). These formulas are very limited in scope

ignoring very important linguistic elements that determine

22

readability of text. According to Rude (1986, p. 129), these
formulas ignore "factors such as figurative use of language,
technical terms, sentence structure, the coherence of topics,
and connectives used between clauses and sentences”. When
determining difficulty, textual features, such as coherence
of the ideas in a text and their density and connectives,
also need to be considered (Singer and Donlan, 1985, p. 195).
Also, reader-related factors need to be considered in
assessing difficulty, such as ”the reader's prior knowledge,
vocabulary ability, reasoning processes, purposes, and goals
in reading the text" (ibid.). Reading difficulty depends on
factors both in the reader and the text. According to Singer
and Donlan (ibid.), "reading difficulty for a particular
individual depends upon an interaction between the text and
the individual."

Despite the limitations of readability formulas, they
have one strength in their favor, their stress on the
importance of vocabulary in determining the level of
readability. Coady (1987, p. 102) points out that "in
attempting to measure the readability or difficulty of a
given text, it has long been known that vocabulary load is
the most significant predictor." To reinforce this, Coady
quotes Chall (1958, p. 157) who says that "once a vocabulary
measure is included in a prediction formula, sentence

structure does not add very much to the prediction."

23

Sentence structure certainly does affect readability, but the
readability formulas as given above do not include data on
the internal structure of sentences, including the depth of
embedding involved, so with respect to grammatical structure
their results seem to be rather superficial. Sentence length
can provide only a rough estimate of grammatical difficulty.
Research on the effects of syntactic variables on readability
is an area that needs further research (Coady, 1987, p. 103).
As a result, we cannot be dogmatic.

As far as the contribution of linguistics to reading is
concerned, there is a hierarchy of several different levels
in operation during reading: ”textual, syntactic, morphemic,
lexical, graphophonemic, or graphemic, which have been
experimentally evidenced in reading" (Ulijn, 1984, p. 67).
The question is how do all these factors work together to
make reading possible? The way that Ulijn (ibid., p. 68)
would handle this "is to suppose an interaction between all
linguistic levels in reading; i.e., that they are operating
without a strict bottom-up or top-down order". In order to
sort out this interaction, "even though there is interaction
among these linguistic factors, there is the need to isolate
the individual factors which contribute to problems with
reading comprehension (Strother and Ulijn, 1987, p. 91).

The assumption behind readability formulas is that

readability is determined by word frequency. According to

24

Anderson and Freebody (1981, p. 77), "an assessment of the
number of meanings a reader knows enables a remarkably
accurate prediction of this individual's ability to
comprehend discourse." Furthermore, they add that ”an
equally consistent finding has been that word knowledge is
strongly related to reading comprehension” (ibid., p. 78).
The authors cite a number of studies to illustrate the
importance of vocabulary in reading comprehension. In a
study by Thorndike (1973), data was collected on at least
100,000 students in 15 countries and grouped into three age
groups. According to Anderson and Freebody (1981, p. 78),
Thorndike "found median correlations between vocabulary
knowledge and reading comprehension, corrected for test
reliability, of .71 (lo-year-olds), .75 (14-year-olds), and
.66 (17-18-year olds)." Anderson and Freebody (1981, p. 78)
also cited a study by Coleman (1971) on reading difficulty of
prose. Coleman (1971, p. 184) found that although sentence
complexity had some importance as a variable, ”word
complexity (number of letters, morphemes, or syllables;
frequency of usage) will account for about 80 percent of the
predicted variance”. Finally, Anderson and Freebody (1981,
page 80) cited a study by Klare (1974-1975) in saying that an
adequate readability formula need only have two variables,

one of word difficulty and one of syntactic difficulty.

25

Readability formulas are highly dependent on word
frequency in determining the degree of comprehensibility of a
passage. In a study by Marks, Doctorow, and Wittrock (1974),
increased use of high-frequency words was shown to improve
reading comprehension. Two different versions of five
stories were given to sixth-grade children. In the two
versions, 15% of the words were altered in frequency with one
version containing high-frequency and the other low-frequency
equivalents. The frequency counts were obtained from
Carroll, Davies, and Richman (1971). The authors summarized
the results of the experiment as follows:

...the magnitude of the increase in reading

comprehension due to the 15 percent increase in high

frequency words was marked. The average increase in
reading comprehension from the less meaningful to the
more meaningful treatment across all comprehension tests
at all reading levels was 25 percent...(Marks,

Doctorow, and Wittrock, 1974, pp. 261-262).

Such a study shows the importance of word frequency in
readability formulas. However, these formulas are as
accurate as the word frequency counts. In referring to the
dependence of readability formulas on word frequency counts,
Wardhaugh (1969, p. 88) warns that "any research based on the
use of the formulae may be invalid if the counts themselves
are inadequate". Word frequency needs to be balanced with

other considerations, such as linguistic and rhetorical

structure 0

26

A further study investigating the effect of difficult
vocabulary on comprehension as measured by 1) free recall, 2)
summary recall, and 3) sentence recognition was conducted by
Freebody and Anderson (1981). For the first experiment, the
effect of difficult vocabulary was studied under three
experimental conditions:

The ”easy" condition comprised the high word-frequency

form of the passage; the "medium" condition entailed the

substitution of approximately 1 substance word in 6 in
the easy condition with a low-frequency synonym; the

"difficult" condition entailed such substitutions for 1

substance word in 3 (ibid., p. 7).

The study used 105 sixth-grade students in order to study
five passages under the three experimental conditions. In a
surprise result in this experiment, it took a high density of
low-frequency vocabulary to achieve a deterioration in
performance. The authors (ibid., p. 27) said that ”only when
one substance word in three was changed to a low-frequency
synonym did performance deteriorate reliably across the
passages used in Experiment 1". For the second experiment,
”difficult vocabulary was placed in important text elements
in one form of the passages, and in unimportant elements in
another" (ibid., p. 1). The effect in comprehension was only
seen in the summary measure. The authors explained the
results of this experiment by use of the "minimal effort
principle" where the reader spends as little effort as

possible on the processing of the text as long as the main

themes are understood. However, if difficult vocabulary is

27

found in important propositions in a text, these words must
be processed, and if they are not understood, understanding
of the text would be impeded. If difficult vocabulary occurs
in unimportant propositions, the vocabulary can merely be
skipped with little effect on comprehension. In summary,
readers use strategies to avoid difficult words.
2.3. Culture and Reading Comprehension. Barnitz (1986) did
a literature survey on the relationship between culture as
expressed in content schemata and second language reading
comprehension. Historically, an early study of the effects
of cultural schemata was conducted by Bartlett (1932). When
subjects from England in the experiment read American Indian
folk stories, the effects of processing text based on
different cultural traditions were evident. Later both a
study by Kintsch and Green (1978) who had college students in
America read stories in the Apache folk tradition and stories
written by Grimm and a study by Steffenson, Joag-Dev, and
Anderson (1979) who had students in American and Indian
universities read letters about weddings in India and America
further corroborated the findings of Bartlett through more
sophisticated studies which showed that cultural schemata
affected comprehension (Barnitz, 1986, p. 99).

Barnitz (1986, p. 101) discusses a study by Johnson
(1981) which examined the effect of cultural schemata on

reading English as a second language. The subjects were

28

intermediate and advanced students who were studying English
as a Second Language and students who were native speakers.
They read two versions of American and Iranian folk stories:
one had simple syntax and the other had complicated syntax.
One of the conclusions was that syntax did not have as
important an effect on readability for the Iranian subjects
as did the cultural background of the text. For the American
subjects, readability was influenced both by the grammatical
difficulty and culture assumed by the text. Finally, the
subjects in both groups could reason better in recollecting
text with a cultural background that matches their own.

Johnson (1982) conducted a further experiment on the

importance of background knowledge. The subjects were 72 ESL
students at the advanced level. They read a passage about
Halloween containing both familiar and unfamiliar information
(Johnson, 1982, p» 503). There were four treatment
conditions (ibid., p. 507):

Group 1: reading the passage without a vocabulary list
to study before reading or to refer to while
reading.

Group 2: studying the definitions of the target words
before reading but not being able to refer to
this list while reading.

Group 3: reading the passage with the target words
glossed in the passage.

Group 4: studying the target vocabulary words before
reading with the definitions of the target
words glossed in the passage.

The subjects were asked to recall the story and to recognize

information about the story. According to Johnson (ibid., p.

29

508), ”prior experience in the American culture seemed to
affect ESL students' comprehension of a passage on the topic
of an American custom, Halloween". As a result, there was "a
significantly better written recall of the familiar
information in the passage than of the unfamiliar
information” (ibid.).

Barnitz (1986, p. 105) also discussed psychological
processing studies. He noted that there has been little
research "on cross-cultural discourse processing" (ibid.).
He does, however, describe research done by Hinds (1983b):

Hinds conducted research on Japanese and English readers

who read texts in their own language. The experimental

texts in both languages conformed to Japanese rhetorical
patterns. Recall protocols indicated that English
readers processed the Japanese discourse with far more
difficulty than did the Japanese readers. The
differences in Japanese and English discourse patterns
were found to affect the processing strategies of

English readers (Barnitz, 1986, p. 105).

This research indicates that readers use the discourse
patterns of their L1 in processing L2 discourse.

2.4. Computers and Computer-mediated Text. Computer-
mediated text provides manipulations for text in order to
supply help and remediation to the reader while the text is
being processed. Reinking and Schreiner (1985) and Reinking
(1988) investigated the interaction of the reader with
computer-mediated text. These studies have shown positive

effects on reading comprehension of English text which is

computer-mediated .

3O

Reinking and Schreiner (1985, p. 536) described the
potential of the computer-mediated text as follows:

The options for interacting with text displayed on the
printed page are limited by conventional print and the
reader's internal strategies. Hypothetically, computer
technology might influence these processes by
manipulating text in ways not available or feasible in
printed pages.

To test out this potential, 104 fifth-graders and sixth-
graders were presented with 6 passages that were expository
in nature under four experimental conditions. The four
experimental conditions were as follows:

1. Subjects read passages off line (on printed pages).
no textual manipulations were available (off-line
group).

2. Subjects read passages on line (displayed by the
computer). No textual manipulations were available
(test-only group).

3. Subjects read passages on line and were free to
choose from among several textual manipulations
which were deemed useful for enhancing comprehension
of the passages (select-options group).

4. Subjects read passages on line and then were
required by the computer program to view all
available manipulations before being allowed to
continue (all-options group) (ibid., p. 540).

An analysis of variance showed "that computer-mediated text
can influence reading comprehension and that comprehension
was most consistently increased when manipulations of the
text were under computer control" (ibid., p. 536).

Reinking (1988) did a further study to confirm the
results of the previous study. He employed the same
(experimental conditions as in the previous study. Along with

confirming that computer-mediated text actually improved

3]

reading comprehension, the purpose of this study was to
measure the time taken to read the texts and to determine
whether that affected the comprehension. Further, Reinking
(1988, p. 484) wanted to learn whether computer-mediated text
”would affect passage preference" and the estimation of
learning on the part of the subjects. Thirty-three fifth-
grade and sixth-grade students read the passages employing
the same options as the previous study (ibid., p. 488). They
were also asked two additional questions after they had
completed the reading (ibid., p. 489):

1. How much did you like this passage?
2. How much did you learn from reading this passage?

The result of this study was that the conclusion of the
previous study was confirmed. The comprehension of the
subjects using computer-mediated text increased. Also, the
subjects took more time to read computer-mediated text.
However, "even after the effect of reading time was removed
statistically, comprehension scores remained significantly
higher for readers of the computer-mediated texts that
offered computer assistance" (ibid. p. 484). The response to
questions 1 and 2 above indicated that the computer
presentation of text did not affect the passage preference
for the readers nor their estimation of the amount of
learning they felt they had achieved (ibid. pp. 493-494).

One way computer-mediated text can be presented to

learners is through hypertext. Hypertext itself refers to

32

"non-sequential text" (Last, 1989, p. 146). Hypertext allows
the user to choose a word or larger unit on the screen and
have the program define, translate, or explain it. In a
hypertext-based program, a database is created where each
entry is called a node and contains at most one screenful of
data or text. Non-sequential associative links between nodes
allow the user to go in any direction. This process is
referred to as navigation. Since orientation is one of the
problems with this system, a utility, called a browser, has
been developed to tell the user where he or she has been,
what the current positions is, and what the options are for
the next move. In this way, a lesson can be created by
piecing it together in any way desired. When audio, video,
and graphics are interfaced with hypertext, then it becomes
hypermedia so that images and sound can be used to reinforce
text. Adding a mouse makes the system even easier to
interact with. The student can move through the learning
material at the pace and ordering of his or her own choosing.
This kind of set-up is ideal for presenting reading materials
because the student can control the rate of his or her own
learning. The student can choose the order in which material
is covered. If the student is having trouble in
understanding certain words or sections of the text, he or
she can use the mouse to click these sections on the screen

.in order to receive definitions, explanations, and other

33

forms of help. The hypertext system is designed to give the
student power to choose what help is appropriate and needed
to understand textual material. Hypertext can provide all
the help needed to understand a text in a convenient and
easily-accessible form.

An example of a hypermedia approach to learning Chinese
characters used in Japanese is KanjiCard developed by Kazuko
Nakajima (1988). The program uses Hypercard, Apple's
implementation of a hypermedia environment. KanjiCard uses
text, graphics, animation, and sound in a self—tutorial
program to teach the use of 300 Chinese characters in
Japanese. According to Nakajima (1988, p. 80) in reference
to her program, "a variety of information on individual Kanji
is organized hierarchically into four major categories: (1)
graphic, (2) sound, (3) semantic, and (4) usage.” A user can
access this information by utilizing the directory which is a
map of all the information that is available. When the user
chooses the graphics section on the menu, a 2-inch by 2-inch
image of the Chinese character is drawn (ibid., p 81). An
option to draw the stroke order shows the Chinese character
drawn stroke—by-stroke. When the sound section is chosen,
the user can hear the Chinese character pronounced (ibid. p.
82). The section on semantics provides information as to the
origin of the Chinese character and discusses its meaning

(ibid.). Finally, when the usage section is chosen, the user

34

gets information on Chinese compounds and related Chinese
characters (ibid., p. 83). Furthermore, example sentences
are given to illustrate the use of a particular Chinese
character in context. The main advantage of this program is
that a lot of information related to Chinese characters is
readily available in an easily accessible format. Learning
is accomplished in an interactive format through the
integration of sound, graphics, and text.

On February 3, 1991, Thurn (1991) gave a demonstration
at The Third Conference of the Lower Lake Erie Teachers of
Japanese at the University of Pittsburgh. The demonstration
featured an introduction to Interactive Japanese:
understanding written Japanese, a program that teaches the
reading of Japanese using a hypertext environment. The
program provides help for the user with the aid of an on-line
dictionary, access to information on words, phrases, and
sentences, and provision of background information on the
text.

2.5. Japanese Textual Variables. Grabe (1987, p. 118) in
his attempt to define expository prose as a discourse genre
used a variety of grammatical variables which were then
counted to discriminate expository text types. The text
'variables were as follows (ibid.):

Syntactic Variables

1. Prepositions (by, of, with...)
2. Nominalizations (-tion, -ity, -ment...)

35

3. lst/an person pronouns

4. 3rd singular pronoun

5. Singular pro-verb DO

6. Past tense

7. Present tense

8. WOrds per sentence

9. Adjectives

10. General conjuncts (and, but, or...)

11. Precise conjuncts (namely, thus...)

12. Subordinators (because, since...)

13. Passive

14. Infinitives

15. WH-clauses (following verbs like tell, remember...)
16. THAT-clauses (following verbs like know, say...)
17. Relative clauses.

18. Pied piping (prep. relative cl.)

19. General hedges (maybe, around...)

20. THAT-deletion (in complement clauses)

21. Relative pronoun deletion

22. Contractions (aren't. . .)

23. All attitude adverbs (surely, truly...)

24. All clefting constructions

25. Split auxiliary/infinitives

26. All questions

27. All locative adverbs (e.g., of time, place)

Cohesion Variables

28. Definite article reference

29. Deictic reference

30. All repetition

31. Lexical inclusion

32. Lexical comparatives

33. Lexical synonymy/antonymy
This present study used a parallel set of textual variables
for Japanese based upon the syntactic analyses of Japanese of
Alfonso (1966), Martin (1975), and. Kuno (1973), the
rhetorical analysis of Hinds (1983a, 1983b, 1987, 1990) and
Grabe and Kaplan (1989), the lexical analysis of Lado (1972),
and the cultural analysis of Barnitz (1986). This parallel
list of textual variables for Japanese is used to measure

textual difficulty in the reading of Japanese.

36

2.6. The Gap in Japanese L2 Reading Research. There is a
gap in the reading research of Japanese as a second language
'with respect to the effect that textual variables have on the
readability of Japanese text (for a discussion of the textual
variables for Japanese used in this study, see Chapter 3,
section 3.1). Although I have found no study in this area, I
think that it is an area that certainly needs to be explored.
Even in the area of L2 English reading comprehension, Coady
(1987, p. 103) states that more research is needed on how
individual syntactic variables, such. as passives and
nominalizations, affect comprehension. The same is certainly
true for L2 reading of Japanese. The main purpose here is to
study the effect of syntactic variables on L2 reading of
Japanese. The following are the hypotheses that this present
study will address:
1. Grammatical variables affect the difficulty of the
reading of a Japanese text by native speakers of
English.

2. Computer-mediated text can facilitate the
comprehension of a passage more than ordinary text.

37

CHAPTER 3
Method

3.0. Purpose. The purpose of this dissertation is two—fold:

1) firstly to assess difficulty in Japanese text in
relation to reading comprehension.

2) secondly to show that computer software that has

been designed to meet the expected difficulty in a

Japanese text can facilitate reading comprehension.
In order to fulfill this dual purpose, this pilot study
focuses on the relationship of grammatical structure to
passage difficulty. Along the way, the effect of rhetorical
structure, vocabulary, and culture on textual difficulty is
also considered in order to fill out the total picture of the
major factors that affect the level of difficulty of Japanese
text. All these factors help determine the burden that faces
the reader of Japanese as a second language. In this study,
computer software is designed around these factors in order
to determine 1) whether it can assist in measuring
quantitatively the contribution these factors make to passage
difficulty and 2) whether it can actually facilitate the
reading of Japanese text. This software is tested on human
subjects in order to measure its effectiveness in measuring

textual difficulty and its power to facilitate the

comprehension of Japanese written text.

38

3.1. Variables in this Study. This dissertation considers a
list of 20 categories as possible variables in passage
difficulty. This list is really an enumeration of 20
possible variables that are hypothesized to influence passage
difficulty. These variables are mainly syntactic in nature
and are the main focus of this study; however, consideration
is also given to variables that are rhetorical, lexical, and
cultural in nature. The following are the variables which
are hypothesized to affect passage difficulty (Alfonso, 1966;
Grabe, 1987, p. 118; Kuno, 1973; Martin, 1975: as discussed
in chapter 2, section 2.4):

film-

1. Homophonous particles.

a. ni marking benefit, goal, locative, purpose, result,
the agent, or the indirect object.

b. to marking the conjunctive or the comitative.
c.‘ o'marking the direct object or the place traversed.

d. no indicating the locative genitive or the objective
genitive.

e. .mo meaning 'also', 'even', or 'as much as'.

2. Nominalizers -- koto and no.
3. Anaphora.

4. Tense.

5. WOrds per sentence.

6. Noun head modification.

7. Conjunctions.

39

8. Subordination.

9. Passive.

10. Causative.

11. Verb complexes -- verb compounding.

12. Quotations -- Quoted material comes before the quotation
particle to.

13. Relative clauses.
14. Hedges.

a. kamoshirenai 'maybe'.

b. daroo 'perhaps, probably'.
15. Further embeddings.
16. Comparatives.
Discourse-
17. Rhetorical structures.
mamm-
18. The percentage of kanji in a passage.
19 . Vocabulary.

a. Formsmeaning relationships between English and
Japanese.

1) WOrds similar in form and meaning.
2) Words similar in form but different in meaning.
3) WOrds different in form but similar in meaning.
b. Idioms.
Culina-

20. Cultural assumptions.

4O

Items 1 through 16 relate to grammatical structure in
keeping with the primary hypothesis of this dissertation:

Syntactic variables affect the difficulty of the reading
of a Japanese text by native speakers of English.

In particular, embedding, case marking, and other forms of
sentence complexity are considered in the determination of
passage difficulty.
3.2. The Evaluation of Difficulty in Two Texts. Two
Japanese texts are evaluated from the viewpoint of the effect
of grammatical, rhetorical, lexical, and cultural features on
textual difficulty. Most of the features used in the
evaluation are grammatical in nature, so the constructions in
these passages which contribute to grammatical complexity are
focused on in order to illustrate some of the issues involved
in passage difficulty. Each of the features involves a
specific hypothesis as to what contributes to passage
difficulty in reading comprehension. Specific examples in
these passages that illustrate the relation between these
hypotheses and passage difficulty are considered. After the
two passages are considered in the light of the 20 variables,
the difficulty level of each passage is calculated to
determine which of the two is more difficult.

The two passages are chosen so that neither is too
difficult for the subjects. Yet there needs to be a clear
distinction in the difficulty level of these two passages in

order to test the validity of the 20 variables as features of

41

textual difficulty. The more difficult passage is an article
that originally came from the Asahi_§hinbun newspaper column,
called "Tensei Jingo". This newspaper also publishes an
evening English newspaper, called The_A§ahi_Eg§ning_N§g§, in
which English translations of the articles appear. In The
W, a translation of the "Tensei Jingo"
column appears, which is called "Vox Dei, Vox Populi". The
article used in this study, called "Aruke Orinpikku"
(translated “The Walking Olympics"), appears in a collection
of these articles published by the Asani_§ninbgn editorial
staff (Asahi Shinbun Ronsetsu Iinshitsu, 1985, pp. 40—41).
The easier passage is taken from an elementary school reader
for the 4th grade, called WM (Ishimori,
1977, pp. 42-43), which is translated as 'Elementary New
Japanese'. This passage is still quite challenging for
university-level readers of Japanese as a second language.
Any selection of reading passages must take into account the
reading level of the subjects of the experiment. With the
on-line help provided by computer-meditated text, these
readings should be within the reading capabilities of the
subjects.

3.3. An Experiment Assessing Difficulty in the Two Texts.
These two Japanese texts, one rated difficult and the other
easy with regard to the 20 variables, were investigated with

respect to the actual difficulty they present to English

42

speaking students who are reading Japanese as a second
language. The readers in this study were presented with
these texts and were asked to circle or underline the parts
of the texts they thought were difficult. The purpose of
this task was to critique the validity of the assessment of
difficulty presumed by the 20 criteria as listed above. If
the students could clearly indicate that, in fact, one text
is easier than the other, then the validity of the criteria
as indicators of textual difficulty would be indirectly
confirmed. The actual analysis of the validity of the
predictive ability of 20 variables was based on two
calculations: 1) the percentage of predictions that were
correct and 2) the agtgal problem areas as determined by this
experiment that were successfully predicted. Furthermore,
the determination of which passage was more difficult was a
function of the number of total marks in that passage. These
calculations are further explained in chapter 5.

The subjects that participated in this study were
members of intermediate level Japanese classes at a large
Midwestern University. They were 3rd- and 4th-year students
who had studied the fundamentals of grammatical structure and
could recognize at least 800 Chinese characters. The fact
that all the subjects were English-speaking has certain
implications for the evaluation of passage difficulty because

clearly the issues of textual difficulty involved in reading

43

Japanese as a second language are related to the L1
background of the reader. These implications are a part of
this pilot study.

3.4. Computer-mediated Text. Based upon both the results of
the experiment assessing the difficulty in both texts and the
linguistic analysis, a computer program was written focusing
on the potential problem areas in the reading comprehension
of these two passages. The program focused on the areas of
difficulty that were indicated by the linguistic analysis and
the experimental evaluation of the performance of the target
population during the experiment assessing the difficulty in
both texts. The computer program provided help in these
areas of anticipated difficulty for the target population.
The goal was to provide the most appropriate help to
intermediate-level Japanese students.

In this computer program, computer-mediated text is
manipulated in a variety of ways. A passage is placed on the
computer screen in a text window which the reader interacts
with. At any point during the reading process, the reader
has certain resources available which can be used to
facilitate understanding of the text. These resources are in
the form of a help menu which presents certain choices to
facilitate comprehension. On the help menu, the reader can
receive help in the following areas: 1) Information on

vocabulary, 2) information on Chinese characters (kanji), 3)

44

information on grammar, 4) information on rhetorical
structure, and 5) information on culture. The grammatical
information given by the help facility is the most complete
since this is the focus of this study. The assistance
provided by this program incorporates the textual analysis
and assessment described in sections 3.2 and 3.3. The user
gets help in the areas indicated as difficult by this
analysis and assessment. These predicted difficult areas in
the text are underlined in the computer-mediated text, so
they can be chosen by the reader in order to provide special
help to understand the text. Since analysis and assessment
were done on the texts before they were presented as
computer-mediated text, it is possible to anticipate certain
problems in the text for reading comprehension so that
remedial explanations can be prepared in advance for these
potential problem areas. These explanations are part of the
facilities available for computer-mediated text. That is,
through the use of textual analysis and assessment described
in sections 3.2 and 3.3, computer-mediated text can be
targeted to a certain group of users in order to give them
the appropriate help they need to comprehend text.

Also in this computer program, the kind of information
the reader uses while doing the lesson is monitored. It
keeps track of the number of times information on vocabulary,

Chinese characters, grammar, rhetoric, and culture is

45

accessed by the reader. An analysis can then be made of this
information over a group of users to determine what was
difficult in the texts. The assumption is that the more
difficult a certain element of text is, the more the users
will need information to explain that element, and therefore,
there is a proportional relationship between the number of
times information about a textual element is accessed and the
difficulty of that element. Thus, difficulty can be
determined by the number of times information on certain
textual elements is accessed. This sort of information was
used to provide an overall picture of the sources of textual
difficulty; thus, the computer can function as a tool for
linguistic study.

3.5. A Test of the Software. The software developed for
this pilot study was assessed under two treatment conditions.
The first treatment condition presented text in conventional
print on a piece of paper, and the second presented text on
the computer screen with the freedom to choose options on the
help menu to aid in comprehension. Students read the two
passages under one of the two treatment conditions. One
group of students read the texts in the off-line treatment
condition (condition 1) and the other group in the on-line
treatment condition (condition 2). Following the passages in
both treatment conditions, there were a series of six

multiple-choice comprehension questions giving four options

46

with three distractors and one correct answer. Without
looking at the passages, the subjects in both treatment
groups answered these questions in order to test their
reading comprehension. Along with keeping track of the
comprehension score, the program kept track of the reading
time from when the subject started to read a passage until he
or she decided to answer the questions. Also, the amount of
time needed to complete the comprehension questions was
recorded. Furthermore, a record was kept of the options
provided by the help menu which were used by the subjects in
the on-line treatment group in order to determine which
options they preferred to use in increasing their knowledge
of the text.

3.6. Summary. This pilot study was carried out to
investigate the causes of difficulty in Japanese texts and
the possible contributions computer-mediated text can make to
the comprehension of text. The method described above
attempted to empirically measure the extent to which
variables of textual difficulty, particularly syntactic
variables, affected the overall difficulty of a text. In
conjunction with this, the role of the computer in
quantifying the contribution of these variables to textual
difficulty and in facilitating the comprehension of text was
considered. This pilot study attempted to confirm the

hypotheses of this investigation on an empirical basis. From

47

this perspective, reliable conclusions can be drawn about the

validity and relevance of these hypotheses.

48

CHAPTER 4

Assessment of Difficulty in Japanese Text

4.0. Introduction. This chapter will assess difficulty in
two Japanese texts according to twenty features. Most of
these features are grammatical in nature and are related to
hypotheses as to what contributes to passage difficulty in
reading comprehension. Each feature will be introduced,
explained with examples from the reading passages "Aruke
Orinpikku" ("Walking Olympics") and "Hoogen to Kyootsuugo”
("Dialects and the Common Language"), and examined in light
of its impact on reading comprehension. These two texts were
examined in light of these features through the use of a
system of quantification where each appearance of a feature
was worth one point (except for feature number 5 as will be
explained later). This point system is just a rough way to
predict difficulty in these passages and is based on the
assumption of a null hypothesis that each feature contributes
an equal amount to passage difficulty. That is, the
assumption is made that the impact of each feature on reading
difficulty is the same. In reality, different problem areas
in a passage may contribute to difficulty in different
degrees. For example, a rhetorical or cultural problem may

Jhave more impact in determining reading difficulty than a

49

grammatical problem. The assignment of a uniform one point
to each problem area is rough estimate of each problem's
contribution to total textual difficulty. The assumption of
the null hypothesis is made in order to begin the research
and certainly evens the features out too much in terms of
impact on difficulty but does constitute an approach to make
some rough estimates of reading difficulty. In future
research, the focus should be on the relative contribution of
each feature to overall passage difficulty. All of the
points that were accumulated in evalulating all the sentences
in a passage were added up, and to that was added the average
sentence sentence length to create an estimate of passage
difficulty. This estimate was then divided by the number of
words in the passage in order to factor in passage length

into the readability estimation. The formula is as follows:

passage difficulty points

 

readability =
passage length in number of words

This readability score is an estimate of the difficulty in
comprehending a reading passage. It is only meant as a
general heuristic of reading passage difficulty to guide in
the estimation of readability. Later in chapter five, a
report will be given on how these features were empirically

tested as accurate predictors of passage difficulty.

50

4.1. Features Counted in the Estimation of Reading
Difficulty. The following is a discussion of each of the
features which contribute to reading difficulty. Each
feature is explained and accompanying examples are provided.
4.1.1. Homophonous Particles.
4.1.1.1. The Particle ni. The particle ni can mark the
dative, locative, or the agent along with other cases
(Martin, 1988, pp. 40-41). This homophony can be a potential
source of confusion. An example of the use of hi to mark the
agent of a passive would be the following:

(1) Jenn ga sensei p; homeraremashita.

'John was praised by the teacher.‘

In this sentence, ni marks the agent sensei 'teacher'.
However, in a passive sentence, the particle ni may not mark
the agent and may be a source of confusion. An example of
this is sentence number 6 from "Aruke Orinpikku" (see
Appendix A.for the passage):

(2) YOkkakan arukitooshita mono ‘pi kanpo

Four days walk-continued people to completion
of walking

no shoosho to medaru ga

possessive certificate and medal nom
particle particle

ataerareru dake da.
be given only are

'It's just that a certificate of completion and a
medal are given to those who last the distance over

51

four days.‘

The phrase YOkkakan arukitooshita mono hi 'to those who walk

four days' functions as the dative of benefit and not as the

agent with respect to the verb.

Sentence 2 of "Hoogen to Kyootsuugo" ("Dialects and the

Common Language") is a further example of this sort of

confusion.

(3) Nihon no chisei wa yama ya

Japan possessive typography topic mountains and
particle particle
kawa ga ooku, ,motomoto kootsuu
rivers nom. a lot of formerly transportation
particle
31_ fuben de aru ue, .mukashi
dative of inconvenient be in addition in
reference to previous
times

we, nihyaku suujuu mo no
contrastive 200 10, 20, 30, etc. attributive
particle (approximately) genitive
daimyoo no ryoochi n1
feudal possessive property, particle indicating
lord particle fief dative of benefit
wakarete seiji ga okonawarete ita.
be divided administration nom. was being carried

particle out

'As for Japan's physical geography, in addition to
there being a lot of mountains and rivers naturally
making it difficult for transportation, in former
times, Japan was divided into 200 to 300 fiefs,
owned and administered by feudal lords.‘

52

In

The phrase nihyaku suujuu mo no daimyoo no ryoochi ni 'into
200 to 300 fiefs owned by feudal lords' does not indicate the
agent with respect to the passive verb wakarete 'be divided'
but the result.

This brings us to an important question: how should
points be awarded to indicate passage difficulty based upon
the potential confusion above? Normally when one comes to a
passive accompanied by the particle ni while reading, the
assumption is made that the agent is indicated since that is
what one would expect. It is, in other words, the unmarked
case. Anything that is unexpected is the marked case and
therefore should cost something. In this case, the cost is
increased difficulty which results in the awarding of a
difficulty point of one. In sentences (2) and (3) above,
since the particle, n1”, does not mark the agent, it is
assigned the value one.
4.1.1.2. The Particle to. Not only can the particle to be
used in the conjunctive and comitative sense (ibid., p. 44),
but also it can indicate an impressionistic modifier. In the
conjunctive sense, this particle functions as a conjunction
indicating ”and”, and in the comitative, it has the meaning
of "with". An example of the conjunctive use of the particle
to'is in sentence 6 of "Aruke Orinpikku" which was discussed
above. The sentence in its unanalyzed form is repeated here

(refer to the analysis above, if necessary):

53

(2) kakakan arukitooshita mono ni kanpo no shoosho £9
medaru ga ataerareru dake da.

'It's just that a certificate of completion apg_a

medal are given to those who last the distance over
four days.’

The particle to joins kanpo no 311003110 'certificate of

completion' and medaru 'medal'.

However, in sentence number 11 of "Hoogen to

Kyootsuugo”, to functions as the comitative:

(4) Nazenara, hoogen wa, sono tochi no seikatsu ya
rekishi.§g fukai tsunagari ga ari, kazokudooshi,
tomodachidboshi, tochi no hitodooshi ga kutsuroida
hanashi o suru toki ni wa,shizen ni kuchi o tsuite

dete kite, tagai ni yoku kimochi o tsuujiau koto ga
dekiru mono dakara de aru.

'The reason for this is a dialect has a close
relation with the life and history of the local
area, so when family, friends, and people of that
region engage in relaxed conversation, the dialect

just rushes out naturally and is able to mutually
convey emotion.’

The analysis of the first part of the sentence is as follows:

Nazenara, hoogen we, sono tochi no

because dialect topic that region possessive
particle genitive

seikatsu ya rekishi £2, fukai tsunagari ga ari

life and history with deep connection nom. exists

part.

The phrase sono tochi no seikatsu ya rekishi Q fukai
tsunagari 'a deep connection with the life and history of

that region' contains the particle to which expresses the

loomitative meaning, 'with'.

54

An example of the use of to to indicate an

impressionistic modifier is sentence 6 of ”Hoogen to

Kyootsuugo”:
(5) Kennichi de.mo, sono tochitochi no hoogen

today even that each area possessive dialect
genitive

we, mainichi no seikatsu no

topic everyday attributive life locative

particle genitive genitive

naka de ikiiki £9 tsukawarete iru.

middle place of vividly is being used
action
particle

'Even today, the dialect of each area is being used
vividly in the middle of everyday life.‘

The phrase ikiiki to 'vividly' is an impressionistic modifier
of the verb tsukawarete iru 'is being used'. Such a use
resembles the comitative in that the meaning is 'with
vividness'. This use of to in impressionistic modifiers adds
difficulty to a text and thus is worth one point.

Clearly, the comitative form of the particle to is the
marked case, so it receives one point. The conjunctive to
also receives a point but for a different reason relating to
the complexity of the noun head. See feature 6, explained in
section 4.1.6.
4.1.1.3. The Particle 0. Another case of confusion can be
caused by the particle o. In this case, 0 can indicate

either the direct object or the place traversed. The

55

following example of the direct object marking property of o

is number 7 from "Beogen to Kyootsuugo":

(6) Fudan yoku tsukau aisatsu no
ordinarily often use greeting possessive
genitive
kotoba g mite miyoo.

words acc.
particle look let's investigate

'Let's look at words that are ordinarily well-used
for greetings.’

Here fhdan yoku tsukau aisatsu no kotoba 'words that are
ordinarily well-used for greetings' is the object of mite
'see'. This is the expected case, so a difficulty point is
not awarded.

Passage difficulty is increased when the particle 0
indicates the place traversed and not the object of the verb.
This construction is used with verbs of motion. Sentences 15
and 16 of "Aruke Orinpikku" are examples of this usage:

(7) Midori no aka .9, aruku.
green attributive hill place of walk.
gen1t1ve mot1on

particle

'They walked over green hills.’

(8) Uhgazoi no michi g, aruku.
canal-along locative place of walk
genitive road motion
particle

'They walk on roads alongside canals.’

56

The phrases midori no oka Q;'on or over the green hills' and
ungazoi no michi g 'alongside canals' indicate the place
traversed and not the object of the verb, and for that reason
add difficulty to the passage.

Another example of a verb of motion is koeru 'pass over
or beyond', which appears in sentence 3 of WAruke Orinpikku":

(9) ,flimappip_g_ggg;g,sekai kakkoku no hitobito ga kono
machi Neimeehen ni atsumatte, aruku.

'More than 20,000 people from various countries
throughout the world gather in the town of Nijmegen
and walk.’
In the relative clause nimannin o koeru 'pass beyond 20,000
people', nimannin is something being exceeded or passed
beyond. Thus 0 indicates the place traversed, and so adds to
the difficulty of this sentence. As a result, a point is
added here.
4.1.1.4. The Particle no. The particle no, the genitive,
can also give rise to varying interpretations based on the
context, which perhaps can lead to reading difficulty.

Compare the use of the particle no in sentences 1 and 14 of

”Aruke Orinpikku":

(10) Mainen oranda .pg_ chiisa na machi de
every year Holland locative small town at
gen1t1ve
ooki na.moyooshi ga okonawareru.
big event nom.

particle is held

'In a small town in Holland, a big event is held

57

every year.‘

(11)‘Aruku koto pg, daisuki na
walk nominalizer objective like very attributive
genitive much form of the
copulative
hitobito ga atsumatte, aruku koto
people nom.
particle gather walk nominalizer
o tanoshimu.
acc. enjoy
particle

'People who really like to walk gather and enjoy
their walking.‘

The no in sentence (10) is the locative genitive indicating
where the big event is located: oranda no '(in) Holland'.
In sentence (11), the no is the objective genitive indicating
that aruku koto 'walking' is the object of daisuki 'like very
much'. In these two sentences, a point is added for no.
(Nominalized forms like aruku koto will be discussed in
section 4.1.2.)
4.1.1.5. The Particle mo. The particle mo generally means
'also' An example of this meaning is sentence 4 of "Hoogen
to Kyootsuugo":
(13) Sono tame, kotoba .mg, hitotsuhitotsu
for this reason language also each one
no chihoogoto ni wakareru

attributive every region in be divided
genitive

58

yoo ni natta no de aru.
came to be nominalizer be

'For this reason, language, too, came to be divided
into individual geographical areas.‘

In this case, mo is best translated as 'also' or 'even'.
However, mo have another meaning as illustrated in the
sentence 2 of ”Reagan to Kyootsuugo":

(3) Nihon no chisei wa yama ya kawa ga ooku,.motomoto
kootsuu ni fuben de aru ue, mukashi wa, nihyaku
suujuu mg no daimyoo no ryoochi ni wakarete seiji
ga okonawarete ita.

'As for Japan's physical geography, in addition to
there being a lot of mountains and rivers naturally
making it difficult for transportation, in former

former times, Japan was divided into 200 to 300
fiefs owned by feudal lords and administered.‘

In the phrase nihyaku suujuu mg 'about 200, 210, 220, and so
on', mo does not mean 'also' but 'as much as'. This reading
for.mo adds difficulty and is worth 1 point.

4.1.2. Nominalizations.

The nominalizers no and koto are used to nominalize
clauses. Martin (1988, p. 841) defines koto as a "general
nominalization that is abstract, habitual, or remote” and
describes no as ”a specific or definite nominalization that
is single, immediate, concrete, or directly perceivable".
The passage "Aruke Orinpikku" has two sentences (number 14
and number 32 in the passage) with koto:

(11) Aruku koto no daisuki na hitobito ga atsumatte,

g;y&g_59;g o tanoshimu.
'People who really like to walk, gather and enjoy

59

their walking.’

(14) Arukitsuzukeru riyuu wa samazama da ga, aruki no
shinzui wa‘yahari, a;y5g_&g;g o tanoshimu koto
daroo.

'There are various reasons for continuing to walk,
but the essence of walking is in its enjoyment.‘

Both sentences employ the nominalization aruku koto
'walking'. This nominalization contributes to difficulty in
two ways. First of all, the form of the nominalization is
different from that found in English. Secondly, this
nominalization contributes to the embedding process. Because
of the increased difficulty, 1 point is added for koto in
these sentences.

4.1.3. Anaphora.

There are differences in English and Japanese in the
utilization of personal pronouns and demonstrative adjectives
and pronouns. An example that contrasts English usage and
Japanese non-usage of pronouns is the following sequence of
sentences 14, 15, and 16 from "Aruke Orinpikku":

(11) Aruku koto no daisuki na hitobito ga atsmatte, arnkg
.59;g_o tanoshimu.

'People who really like to walk, gather and enjoy
their walking.‘

(7) .Midori no aka 0 aruku.

'They walked over green hills.‘

(8) Uhgazoi no michi o aruku.

'They walk on roads alongside canals.‘

60

In both (7) and (8), the subject aruku koto no daisuki
na hitobito 'people who really like to walk' does not appear
in the Japanese text while it appears as the pronoun "they"
in the English version. The very assumption behind the non-
specification of the subject in Japanese sentences (7) and
(8) is that the subject is the same as that of (11). This
section of the passage is not so difficult since the context
makes the intended meaning very clear. When the context is
not as clear, there can be problems in understanding.

Perhaps, a source of difficulty can be found in the
anaphoric element konna 'this kind of' which is located in

sentence 8 of ”Aruke Orinpikku”:

(15) Sanka kyuukai no Kaneko Tamekazu
participation nine times temporal (name)
genitive

san (nihon aruke aruke kyookaichoo ni wa

Mr. Japan walk walk society-head for topic
particle

gonna omoide ga aru.

this kind of recollection nom. exists
particle

'Tomokazu Kaneko, chairman of the Japan walk-wa1k
Society, who has participated nine times, has the
following recollection.‘
The anaphoric element konna refers to the story which follows
about the argument with the woman. The reader needs to

figure out that konna refers to what follows and not to what

precedes.

6I

The plural form of a demonstrative pronoun can also
refer to previous material in a passage. An example of this
is contained in sentences 8, 9 and 10 of "Hoogen to
Kyootsuugo":

(16) "Kenban we" to iu yoru no aisatsu o, "Oban de
gozaimasu" "Oban ni narimashita" aruiwa, tada "Oban"

to iu tochi ga aru.
'There is a region where the evening greeting

KOnban wa 'Good Evening' becomes Oban de gozaimasu
or Oban ni narimasu.‘

(17) Mata, "Arigotoo" no kawari ni, "Ookini" to ka
"Dandan" o tsukau chihoo ga aru.

"Again, there is a district which uses Ookini and
Dandan instead of Arigatoo 'Thank you.'.

(18) WW ni wa, sono tochi sono

tochi no aji ga kamotte iru.

'As for these words of greeting, they are full of
the flavor of each particular region.‘

In the phrase korera no aisatsu kotoba 'these words of
greeting', korera is the plural form of kore 'this' referring
to the various forms of greeting: Oban de gozaimasu, ookini,
etc. Normally, an inanimate noun, like kotoba, in Japanese
does not pluralize, so the plural suffix -ra is added to a
preceding demonstrative pronoun (Martin, 1988, p. 144). In
this case, kore is pluralized into korera and refers to each
separate greeting. This process adds some difficulty to the

reading process and therefore receives one difficulty point.

62

4.1.4. Tense.
Kuno (1973, p. 261) observed that there is a rule for
tense agreement in English. Consider the following examples:

(19) When I asked him last year, he said, "I'll be
finished soon.

In the reported form, this sentence would be:

(20) When I asked him last year, he said he would be
finished soon.

It would not be:

(21) *When I asked him last year, he said he will be
finished soon.

The same is true for relative clauses. Consider the
following pair of sentences:

(22) John turned off the T.V. the children were watching.

(23) *John turned off the T.V. the children are watching.
Clearly the first sentence is grammatical but the second is
not. According to Kuno (ibid.), such rules for tense
agreement do not apply in Japanese. In Japanese, if the
tense of a relative clause is present, then it is possible
that the action of the verb in this clause takes place at the
same time as that in the main clause (see sentence (9)
below). If the tense of the relative is past, then the
action takes place before that of the main clause (see
sentence 2 below). Without a knowledge of this basic
relationship in Japanese, sentence comprehension is made more
difficult. An example of this is sentence 6 of "Aruke

Orinpikku”:

63

(2) kakakan arukitooshita mono ni kanpo no shoosho
to medaru ga ataerareru dake da.

'It's just that a certificate of completion and a
medal are given to those who last the distance over
four days.’
The relative clause yokkakan arukitooshita 'continued to walk
for four days' has its verb arukitooshita in the past tense,
which means the medal and certificate are awarded after
completing the event.

Another example is sentence 3 of the same article:

(9) Nimannin o koeru sekai kakkoku no hitobito ga kono
machi Neimeehen ni atsumatte, aruku.

'More than 20,000 people from various countries
throughout the world gather in the town of Nijmegen
and walk.‘
The relative clause nimannin o koeru 'exceed 20,000 people'
has its verb koeru 'exceed' in the present tense which means
20,000 are gathered for the walk while it is taking place.
4.1.5. WOrds per Sentence. }

This is a rather superficial measurement of sentence
difficulty since it does not take into account complexity in
grammar or vocabulary. However, it is a correct
generalization to say that complex sentences tend to be long,
so perhaps sentence length in itself can be used as a
heuristic guide to grammatical complexity. This is not to
say that it is an infallible guide but can only serve as a

rough approximation of difficulty. In order to count words

in Japanese, judgments have to be made on where to determine

64

word boundaries. Japanese sentences are normally written
with no spaces between the words. In this dissertation, the
words will be divided up according to the rules found in
Martin (1988, pp. 16-17). When these passages are divided up
according to these conventions, the average number of words
per sentence in "Aruke Orinpikku" is 9.97 while in "Reagan to
Kyootsuugo" it is 21.27 (see section 4.3).

4.1.6. Noun Head Modification.

This section covers the modification of nouns by other
nouns and adjectives. There are differences in noun
modification between English and Japanese. In Japanese,
modifiers precede the noun head they modify while in English
they come either in front of (adjectives, demonstrative
adjectives, etc.) or after (relative clauses) the head.
Consider sentence number 1 of "Aruke Orinpikku" as an
example:

(10) Mainen oranda no chiisa na machi de ooki na moyooshi
ga okonawareru.

'In a small town in Holland, a big event is held
every year.‘

The noun phrase oranda no chiisa na machi 'a small town in
Holland' is analyzed as follows: (oranda no) (chiisa na
machi). The particle no indicates the modification of the
noun machi 'town' by the noun oranda 'Holland' and na
indicates that the adjective stem chiisa- 'small' modifies

the noun machi. Here no indicates the locative genitive

65

(Martin, 1988, p. 259). As indicated above, the translation
of this phrase is 'a small town in Holland'. In English, the
head noun 'town' is modified by the adjective 'small' which
comes in front of it and by the prepositional phrase 'in
Holland' which comes after it. This difference in word order
makes it more difficult for English-speaking L2 readers of
Japanese. The modifier in the noun phrase ooki na moyooshi
'big event' is the adjective ooki na 'big'. Here na
indicates that the adjective stem ooki 'big' modifies the
noun moyooshi 'event'.

In the evaluation of difficulty, points are awarded on
the basis of how many modifiers there are. If there is only
one modifier, then no points are awarded. If there are two,
then one point is awarded. If there are three, then two
points are awarded and so on. In the case of modifiers the
general formula for awarding points is:

p = m - 1
where p equals the number of points awarded and m equals the
number of modifiers. The same is true for the coordinating
conjunction to 'and': two or more coordinating conjunctions
‘will be awarded points in the same fashion.

Another example of noun head modification is sentence
number 6 of "Aruke Orinpikku”:

(2) kakakan arukitooshita.mono ni kanpo no shoosho to
‘medaru ga ataerareru dake da.

'It's just that a certificate of completion and a

66

medal are given to those who last the distance over
four days.‘

In the noun phrase kanpo no shoosho to medaru 'a certificate
of completion and a medal', actually two noun phrases are
conjoined by to 'and': kanpo no shoosho 'certificate of
completion' and medaru 'medal'. The nO'which serves to mark
the attributive genitive (ibid.) indicates that kanpo
'completion' modifies shoosho 'certificate'. Note that in
this Japanese phrase, the modifier kanpo no precedes the noun
head shoosho and that in the corresponding English phrase 'a
certificate of completion', the modifier follows the noun
head. This difference adds difficulty to the task that an
English-speaking reader of this Japanese text has in
comprehending it.

Nouns can also be conjoined without the use of the
connective to. An example is sentence 28 of ”Aruke
Orinpikku”:

(24) "Walk" to iu senmon zasshi mo dete iru.

'There is now a special magazine, "walk".'
The noun phrase senmon zasshi 'specialty magazine' combines
two nouns without the connector to. This fact needs to be
taken into account when considering difficulty.
4.1.7. Conjunctions.

This section is a consideration of sentence level
conjunctions which connect two or more sentences on an equal

basis. Because of this, one sentence is not seen as embedded

67

in the other. These conjunctions can be either coordinative
(and), sequential (then), disjunctive (or), consequential
(therefore), adversative (but), corrective (on the contrary),
or explanatory (that is) (Martin, 1988, pp. 817-819).

An example of a sentence-level conjunction is contained
in sentence number 10 of "Aruke Orinpikku":

(25) Suru to, nenpai no josei ga susumidete itta.

'Then an old woman came forward.‘

The sequential connective is suru to 'then' connecting
sentence 10 with 9:

(26) Hajimete sanka shite toki, nakama to issho ni
"faito, faito" to kowadaka ni sakende aruita.

'The first time he participate, he shouted with
other participating Japanese, "Fight! Fightl"'

This connective adds difficulty to the understanding process
because the reader needs to be aware of what happened in the
previous sentence and relate it to the present one.

An example of a consequential connective is sentence 3
in "Reagan to Kyootsuugo":

(27) Sore de, hitobito wa, tagai ni jiyuu ni ikiki suru
koto galmuzukashikatta.

'So it was difficult for people to freely associate
with each other.‘

trhe connective is sore de 'so' relating this sentence to the
jprevious, sentence 2 of this text:
(3) Nihon no chisei wa yama ya kawa ga ooku,.motomoto
kootsuu ni fuben de aru ue, mukashi we, nihyaku

suujuu mo no daimyoo no ryoochi ni wakarete seiji
ga okonawarete ita.

68

'As for Japan's physical geography, in addition to

there being a lot of mountains and rivers naturally

making it difficult for transportation, in former

former times, Japan was divided into 200 to 300

fiefs owned by feudal lords and administered.‘

Sentence 13 of "Aruke Orinpikku" contains an example of

an adversative particle which serves as a sentence-level
connective:

(28) Ganbarou to iikikaseru karui chooshi no "faito"
datta ga, Kaneko san we "we don't want to fight"
to in kotoba ni komerareta mono o sasshite, fukaku
atama o sageta.

'It was a light "Fight!" calling on people to exert
an effort, bu; Kaneko realized the thinking behind
the words, "we don't want to fight,” and hung his
head in shame.‘
The particle ga 'but' connects both parts of the sentence,
and it is up to the reader to interpret the two parts in a
way that preserves the contrast.

Each of these connectives is worth 1 point since the
reader has to relate different parts of the reading to make
sense of the passage.

4.1.8. Subordinators.

Subordinators are used to embed one or more clauses in a
sentence. Examples of subordinators would be : no ni
'although', kara 'because', -te kara 'after', no do
'because', be 'if', tame ni 'for the purpose of', to ”when',
and -nagara 'while'. An example of embedding using the

subordinating connector, te, is sentence 3 of "Aruke

Orinpikku":

69

(9) Nimannin o koeru sekai kakkoku no hitobito ga kono
.machi Neimeehen ni atsumatte, aruku.

'More than 20,000 people from various countries
throughout the world gather in the town of Nijmegen
and walk.‘

This sentence has the following constituent structure:
NP /\

kakkoku no hitobito ga S V

'people from.various countries
countries throughout the world'

kono machi Neimeehen aruku
ni atsumatte 'walk'
'gather in the town
of Nijmegen

 

 

The verb of the embedded clause is in the gerundive form,
atsumatte, of the verb atsumaru 'gather'. Certainly an
embedding pattern like this adds more difficulty to the
reading process because an embedded clause must be processed
within the matrix sentence.

An example of the subordinator no de 'because' is shown
by sentence 26 of ”Aruke Orinpikku":

(29) Mainen no koto na no de, juumin to sankasha wa
en de musubareru to in.

'Since it is an annual event, (they) say that the

residents and participants are strongly tied
together.‘

70

The clause mainen no koto na no do 'since it is an annual
event' is embedded in the sentence thus making it a more
involved structure to process.

Another example of the use of a subordinator is sentence
number 19 from the same passage:

(30) Gunka o utainagara heitai ga aruku.

'Soldiers walk.while singing military songs.‘

In the clause gunka o utainagara 'singing military songs',
the stem form utai of the verb utau 'sing' is combined with
the subordinator -nagara. This subordination causes the
reader to do some extra processing.

Another structure involving verb sequencing brings about
a form of embedding through subordination. Examples of verb
sequencing are: -te kuru 'motion towards the speaker', -te
iku 'motion away from the speaker', -te miru 'try and find
out (something)', -te snimau 'to finish (something)', -te
morau 'the speaker receives (the favor of doing something)
from an equal or inferior', -te itadaku 'the speaker or
someone who belongs to the speaker receives (the favor of
doing something) from a superior', -te kureru 'something is
done for the speaker or someone who belongs to the speaker by
someone equal or inferior', -te kudasaru 'something is done
for the speaker or someone who belongs to the speaker by
someone superior', -te ageru 'the speaker gives (does the

favor of doing something) to an equal or superior', and -te

7l

(yarn 'the speaker gives (does the favor of doing something)
to an inferior'.

The meanings of the verb sequences -te kuru and -te iku
vary according to the situation and the point of view of the
speaker. Sentence 18 from "Aruke Orinpikku" has an example
of this form of sequencing:

(31) "Kiiroi ribon," "Oosuzanna," thshi, moshi kame yo"
to samazama na uta ga kikoete kuru.

'Various songs, such as ”Tie a Yellow Ribbon,” "Oh,
Suzanna," and "Moshi, Moshi Kameyo", can be heard.‘

Ignoring the quoted material, this sentence has the following

constituent structure:

 

 

s v
NP V kuru
”I”, \ I com :
samazama na uta ga kikoete
'various songs' 'can hear'

The gerundive kikoete is embedded within the matrix sentence
which makes processing more difficult in this complex
sentence. The verb kuru expresses the fact that the sound of
the music comes towards the speaker.

This same passage contains an example of the use of -te

kureru 'something is done for the speaker or someone who

72

belongs to the speaker by someone equal or inferior'.
Consider sentence number 25:
(32) KUroobaa no hanataba o okutte kureru shoojo mo iru.

'There are girls who present (the walkers) with
clover bouquets.‘

This clause kuroobaa no hanataba o okutte 'present clover
bouquets' is contained within the relative clause "kurooba no
hanataba o okutte kureru." ('who present (the walkers) with
clover bouquets') which modifies the head noun "shoojo"
('girl'). Again this requires additional processing on the
part of the reader.

All the cases of subordination receive a point because
of the embedding involved.

4.1.9. The Passive.

Case provides further insight into readability. Agents
which are subjects are generally more prominent. The passive
construction acts to reduce this prominence by moving the
agent out of the subject position. Sentence 6 from "Aruke
Orinpikku" illustrates this:

(2) kakakan arukitooshita mono ni kanpo no shoosho to
medaru ga ataerareru dake da.

'It's just that a certificate of completion and a
medal are given to those who last the distance over
four days.‘
The phrases kanpo no shoosho 'a certificate of completion'
and medaru 'medal' are not the agents but the undergoers in

'this sentence. As can be seen, there are no specific agents

73

indicated explicitly in this sentence. As a result, the
agent has been moved out of its expected primary position.
In other words, it has been 'disowned', which can make it
more difficult for the reader to process the sentence. Thus,
a point is added for the passive.
4.1.10. Causative.

The causative construction introduces a new case role,
the instigator (the causer), into a sentence (Martin, 1988,
p. 292). In the case of transitive verbs, the instigated or
the underlying agent (the causee), is marked by the particle
ni (ibid., p. 293). In the case of intransitive verbs, the
causee is marked by the particle o in the coercive case and
by the particle n1 in the non-coercive case (ibid.). This
introduction of a new case role (the instigator or causer)
leads to reading difficulty. An example of a transitive verb

in the causative is the following:

(33) gang ga kodomo ni hoorensoo 0
mother nom. child agent spinach acc.
particle particle
tabesaseta.

be made to eat

'The mother had/made her child eat spinach.’
In this case haha 'mother' is the causer, kodomo 'child' is
the causee, and hoorensoo 'spinach' is the object. An

example involving an intransitive verb is the following:

74

(34) Hang, ga musume o soko e

mother nom. daughter acc. there to
particle particle
ikaseta.

be made to go
'The mother made her daughter go there.’
Here haha 'mother' is the causer and musume 'daughter' is the
causee. Here coercion is expressed. The corresponding non-
coercive of this sentence would be:
(35) flake ga musume ni soko e ikaseta.
'The mother let her daughter go there.’
The causative can also be used to express the permissive
(let...). An example is the following sentence:
(36) Kikasete kudasai.
let hear give
'Let (me) hear.‘
In "Aruke Orinpikku", there is one example of the use of

the causative:

(37) Khroi hada o kagayakaseta musume-sen
tanned skin acc. let shine girls-honorific
particle
ga aruku.
nom. ‘walk
particle

'Girls with glistening tanned skins walk.’

75

The head noun musumesan 'girls' for the relative clause kuroi
hada o kagayakaseta 'making tanned skin shine' serves as the
causer for the corresponding causative sentence:

(38).MUsumesan ga kuroi beds 0 kagayakaseta.

'The girls let their tanned skins shine.’

Kuroi hada functions as the nonvolitional causee for the
sentence. In the causative construction, the nonvolitional
causee is followed by the particle o (Shibatani, 1976, p.
255). This causative construction expresses the permissive.
The real difficulty for the reader is determining the case
relation of musumesan to the relative clause in view of the
fact that the verb of the relative clause is causative.

One point is added when the causative occurs because of
the difficulties it introduces into case roles.
4.1.11. Verb Compounding.

Verb compounding occurs when two verbs are brought
together with the first verb in the stem form. A verb

compound occurs in sentence number 6 of "Aruke Orinpikku":

(2) kakak n azgkigggggiga_mono ni kanpo no shoosho to
.medaru ga ataerareru dake da.

'It's just that a certificate of completion and a
medal are given to those who last the distance over
four days.‘
The verb under consideration is arukitooshita 'continued to
‘walk' which is analyzed as follows:

aruki + tooshita

walk continued

76

The form aruki is the stem form of the verb aruku 'walk'.
The second verb tooshita is the past tense form of the verb
toosu 'continue'. Such a combination can cause difficulty
since the reader has to be familiar with the two verb forms
and how their forms and meanings fit together. As a result,
1 point is added whenever this construction occurs.

4.1.12. Quotations.

Quotations lead to sentence embedding and therefore make
sentence processing more difficult for the reader.
Quotations are marked by the quotation particle to.
Furthermore, quotations need not be direct but can be
reported or indirect speech. An example of indirect speech
is sentence number 11 of "Aruke Orinpikku":

(39) ”Naze tatakaou to in no."

why let's fight quotation say nominalizing
particle particle

"Why do you talk about fighting?"
The embedded clause is tatakaou 'let's fight.‘ with the verb
in the volitional form. The reader needs to recognize the
volitional form in the embedded indirect quotation and
integrate it with the rest of the sentence in order to
comprehend it. The use of the nominalizing particle no'means
that the questioner is asking for an explanation about the
use of the word "fight". The full form of the polite version

of the question would be: Naze tatakaou to in no (desu ka).

77

The form no desu is asking for an explanation from.the person
to whom the question is addressed.

An example of the to in clause used for a direct
quotation is in sentence 28 from the same passage:

(40) "walk" to in senmon zasshi mo dete iru.
'There is now a special-interest magazine, "Walk"'

The clause 'walk' to in 'called walk' defines the name of the
senmon zasshi 'special-interest magazine'. Difficulty is
caused by the reader having to relate 'walk' to senmon zasshi
as the title of the magazine. The clause 'walk' to in ' is a
form of the relative clause (see section 4.1.13 below). The
fact that this clause comes first is the reverse of the
English order which would be: 'a special-interest magazine
called "Walk"'. This reversal of expected word order for
speakers of English adds to the reading difficulty, and 1
point is added. '

4.1.13. Relative Clauses.

Relative clauses are a cause of difficulty in reading
Japanese. There are differences between Japanese and English
in the position and structure of relative clauses. Whereas
in Japanese relative clauses come before the head noun, in
English they come after the head noun. English employs
relative pronouns but Japanese does not. These differences

cause reading difficulties for English-speaking readers of

78

Japanese as a second language. An example of these
differences is sentence number 24 of "Aruke Orinpikku":
(41) Keohii ya kukkii o furumau ie.mo aru.

'There are families which treat (walkers) to coffee
and cookies.‘

 

 

/NP\ VP
Koohii ya kukkii o furumau ie mo aru
'offer coffee and cookies' 'house' 'be'

The relative clause koohii ya kukkii o furumau 'which treat
(walkers) to coffee and cookies' precedes the head noun 19
'house' and does not have a relative pronoun. Also, the
argument in the relative clause corresponding to the head
noun is missing. These differences with English cause
difficulty for English-speaking readers of Japanese.

Another example of a relative clause is found in
sentence 7 of "HOogen to Kyootsuugo":

(42) FUdan yoku tsukau aisatsu no kotoba o mite miyoo.

'Let's look at words that are ordinarily well-used
for greetings.‘

The relative clause is fudan yoku tsukau 'that are ordinarily
well-used' which precedes and modifies the noun phrase

aisatsu no kotoba 'words of greeting'.

79

In conclusion, the fact that relative clauses create
embedding constitutes a source of difficulty for readers.
Additional difficulty is created by the fact that Japanese
relative clauses precede the noun head and have no relative
pronouns to mark their presence. Another source of
difficulty for relative clauses is tense as explained in
section 4.1.4. Thus, 1 point is added whenever a relative
clause occurs.

4.1.14. Hedges.

Sentence 22 in "Aruke Orinpikku" contains a hedge

kamoshirenai 'maybe':

(43) Orinpikku no seishin wa honke yori mo kochira n1

uketsugarete iru no W.

'The real spirit of the Olympic Games may have been

preserved more in this (four days' march) than in

the Olympics itself.’
The verb kamoshirenai adds a hedge to the statement about the
real spirit of the Olympic Games which requires the reader to
do further processing, thus adding to the difficulty.

Another hedge is daroo 'perhaps or probably' which

appears in sentence 32 in the same passage:

(44) Arukitsuzukeru riyuu wa samazama da ga, aruki no
shinzui wa yahari, aruku koto o tanoshimu koto

.QQIQQ-

'There are various reasons for continuing to walk,
but essence of walking is in its enjoyment.‘

80

In this sentence, the author is expressing his opinion with a
certain tentativeness, a nuance which the reader needs to
understand and so 1 point is added.

4.1.15. Further Embeddings.

This classification covers cases not included in the
above. In such cases, a point is added. A further example
of an embedding involves the particle dake 'only'. The
following is an example from FAruke Orinpikku":

(2) kakakan arukitooshita mono n1 kanpo no shoosho to
medaru ga ataerareru dake da.

'I;;§_just_§ha§ a certificate of completion and a
medal are given to those who last the distance over
four days.‘

The embedded clause is YOkkakan arukitooshita mono n1 kanpo
no shoosho to medaru ga ataerareru 'A certificate of
completion and a medal are given to those who last the
distance over four days'. In this case, "dake" follows the
embedded clause and adds a nuance of "so much and no more" to
the action of the clause (Alfonso, 1966, p. 712).

4.1.16. Comparatives.

The way comparisons are structured in Japanese is
different from English. The comparison particle in Japanese
is yori 'than'. An example is the following sentence 22 from
”Aruke Orinpikku" :

(45) Orinpikku no seishin wa honke M kocnira n1
uketsugarete iru no kamoshirenai.

'The real spirit of the Olympic Games may have been
preserved more in this (four days' march) than in

81

the Olympics itself.‘
An analysis of this sentence is as follows:
Orinpikku no seishin wa honke

Olympic possessive spirit topic originator
particle particle

yori mo kochira n1 uketsugarete iru

than this place in has been preserved

no kamoshirenai.

complementizer may be
The phrase honke yori 'than the originator (referring to the
Olympics)‘ is the point of comparison with the "Four Days'
March". The particle mo after yori mo emphasizes the
comparison. In contrasting Japanese with English, yori comes
after what is being compared (in this case, ‘the Olympics‘)
whereas in English "than" comes before it ('than in the
Olympics‘). This difference with English makes the
comparison difficult to read. The occurrence of a
comparative adds 1 point.
4.1.17. Rhetorical Structure.

Discourse constraints also affect the comprehensibility
of a Japanese text. The texts that are being discussed here
are expository in nature. As mentioned in chapter 2, section
2.1., John Hinds (1983a, p. 188) stated that the basic
organization of rhetorical structure for Japanese is ki-shoo-
ten-ketsu where ki means ‘beginning', shoo 'support‘, ten

'change', and ketsu 'conclusion'. The ki begins the topic,

82

the shoo develops it, the ten moves off the track of the
topic to a related area, and the ketsu ties everything
together in the form of a conclusion. However, this is not a
conclusion in the shape of a summary but in terms of a global
statement that brings the argument to a new level. According
to Hinds, this style is neither deductive (a thesis statement
followed by arguments to back it up) nor inductive (a series
of arguments leading up to a thesis statement), but rather
”quasi-inductive” which "has as its purpose the task of
getting readers to think for themselves, to consider the
observations made, and to draw their own conclusions” (1990,
pp. 99-100). A quasi-inductive style in Japanese appears to
lack coherence to an English-speaking reader although to the
Japanese native speaker it is perfectly natural. This point
of view merely assumes that the reader will take a more
active role in getting comprehensible meaning out of a
passage.

The passage "Aruke Orinpikku" exhibits the ki-shoo—ten-
ketsu pattern (refer to Appendix A to see the passage). The
sentence, Mainen, oranda no chiisa na machi de, ooki na
moyooshi ga okonawareru. ‘In a small town in Holland, a big
event is held every year.‘ introduces the topic and begins
the ki. The shoo starts at the third sentence, Nimannin o
koeru sekai kakkoku no hitobito ga kono machi Neimeehen n1

atsumatte, aruku. 'More than 20,000 people from various

83

countries throughout the world gather in the town of Nijmegen
and walk.' and develops the topic further. The ten is
introduced by sentence 27, Ima Nihon dewa, aruki no aikoosha
ga gekizoo shi, yaku kyuunyakuman to no iwarete iru. ‘In
Japan today those who like to walk have increased
dramatically and are said to number about 9 million.‘ The
ten moves the subject of the passage to the popularity of
walking in Japan. Finally the ketsu is expressed by the
concluding sentence, Tanoshiminagara arukeba, kaze no iro ga
miete kuru. 'If you walk while enjoying it, you can see the
color of the wind.‘ This is not the kind of conclusion that
would appear in an English composition, which tends to be a
summary of the topic and its development. The ketsu
functions to move the topic to a whole new level. In this
case, the theme of walking is given a poetic nuance. English
expository writing tends to be organized around a central
thesis with each subtopic supporting the thesis statement and
with the conclusion summing up the argument. English
rhetoric does not employ a ten or an abrupt change (Hinds,
1983a, p. 188). An English-speaking reader while reading a
Japanese expository composition would sense a certain lack of
focus, perhaps making comprehension more difficult. Another
potential troublesome point for English-speaking readers of
Japanese is the conclusion. A poetic nuance would perhaps be

hard to pick up. As can be seen from these observations,

84

Japanese rhetorical structure in itself can create problems
for comprehension on the part of readers who come from
differing rhetorical backgrounds.

This same ki-shoo-ten-ketsu pattern can be found in
"Hoogen to Kyootsuugo" (refer to Appendix A to see the
passage). The ki begins with the first sentence, Nihon wa,
kuni ga semai wari n1, tochi ni yoru hoogen no chigai ga
ookii. ‘For a small country, Japan has large differences in
its geographical dialects.‘ and introduces the topic,
Japanese dialects. The second sentence begins the shoo,
Nihon no chisei wa yama ya kawa ga ooku, motomoto kootsuu ni
fuben de aru ue, mukashi wa, nihyaku suujuu no no daimyoo no
ryoochi n1 wakarete seiji ga okonawarete ita. 'As for
Japan's physical geography, in addition to there being a lot
of mountains and rivers naturally making it difficult for
transportation, in former times, Japan was divided into 200
to 300 fiefs owned by feudal lords and administered.‘ This
sentence begins an explanation of the reason for diversity in
geographical dialects. The ten starts at sentence 6,
Konnichi de mo, sono tocnitochi no hoogen wa, mainichi no
seikatsu no naka de ikiiki to tsukawarete iru. 'Even today,
the dialect of each area is being used vividly in the middle
of everyday life.‘ Here the topic changes to a comparison
between dialects using some language data. The ketsu is

expressed by the final sentence, Nazenara, hoogen wa, sono

85

tochi no seikatsu ya rekishi to fukai tsunagari ga ari,
kazokudooshi tomodachidooshi, tochi 1K) hitodooshi. ga
kutsuroida hanasni o suru toki n1 wa, shizen ni kuchi o
tsuite dete kite, tagai ni yoku kimochi o tsuujiau koto ga
dekiru mono dakara de aru. ‘The reason for this is a dialect
has a close relation with the life and history of the local
area, so when family, friends, and people of that region
engage in relaxed conversation, the dialect just rushes out
naturally and is able to mutually convey emotion.‘ The ketsu
takes the discussion of dialects to a whole new level by
talking about their cultural meanings. In the eyes of
English speakers, this passage could be seen as lacking
cohesion. Again, in Japanese, it is up to the reader to put
everything into a coherent whole.

4.1.18. The Difficult Kanji in a Passage.

Japanese text is written in both kana (hiragana and
katagana) and kanji. If more difficult Chinese characters
are used in a text, the passage becomes more difficult since
readability is directly related to the number of difficult
kanji in a passage. The level of difficulty of Chinese
characters is measured in terms of frequency of use. I am
judging the degree of difficulty of kanji based upon my own
experience as a learner of Japanese. In each of these

passages, each difficult kanji is worth one point. The

86

difficult kanji are underlined in the passages given at the
end of this chapter.
4.1.19. VOcabulary.

Vocabulary is a strong predictor in determining the
difficulty of a passage. Within the domain of vocabulary,
there are several levels of difficulty (Lado, 1972, pp. 275-
288). First of all, words that are very similar in form and
meaning in both English and Japanese are easy to read and
understand. These words enter the language as the result of
borrowing. In particular, words of foreign origin in
Japanese are written in katakana so that words borrowed from
English are easy to spot for English-speaking readers.
Examples from "Aruke Orinpikku" (refer to Appendix A to see
the passage) are orinpikku 'Olympics' in the title, oranda
‘Holland' in sentence 1, foodeezu maachi 'four days' march'
in sentence 2, or koohii 'coffee' and kukkii ‘cookie' in
sentence 24. Secondly, words that are similar in form but
different in meaning in English and Japanese are a little
more difficult but do not constitute major difficulty. An
example of such a word is faito 'fight' which is used in
sentences 9 and 13 in the same passage. For the Japanese who
were walking, saying faito was just a way to encourage one
another not to give up. For Americans, or perhaps even the
Dutch, to use this word while walking would be out of place.

It is certainly something cheerleaders would say during a

87

basketball or football game, but it is not something
participants would say in a walking event. Consequently,
there are differences in the meaning of the word ‘fight' in
Japanese and English. Such differences make understanding a
little more difficult. Thirdly, words that are of normal
difficulty have similar meanings but are different in form.
An example of this would be the word sekai in sentence 3 of
”Aruke Orinpikku". Both sekai and ‘world‘ have similar
meanings, so the reader just needs to learn the Japanese word
form for 'world‘. An example of this from ”Hoogen to
Kyootsuugo” (refer to Appendix A to see the passage) would
be the word semai found in sentence 1. Semai is similar in
meaning to the English word ‘narrow', so all the reader has
to do is remember the form of the Japanese word for 'narrow'.
The above cases range from a low level to a moderate level of
difficulty, so no points are awarded in these cases.

A more serious source of difficulty is the use of
idioms. One example of the idiomatic use of language is
found in sentence 26 of "Aruke Orinpikku":

(46) Mainen no koto na no de, juumin to sankasha wa

W to in-
Since it is an annual event, (they) say that the
residents and participants are strongly tied
together.‘

The phrase in question is en de musubareru 'to be tied by

connection‘, which means the participants are "knit” together

in a common bond. Such an idiomatic usage is very hard for

88

readers of Japanese as a second language to understand, so it
‘will be given one point. Another example of idiomatic usage
is in sentence 21 in the same passage:

(47) Ryukku 0 se ni shita roojin ga aruku.

'An old man with a knapsack on his back walks.‘
The phrase as n1 suru meaning 'put on the back' has the
following literal meaning:

se ni suru

back on do

‘put on the back'

Deriving the idiomatic meaning from this phrase will add
difficulty to the passage, so it is worth one point.

Another source of difficulty is word frequency. Less
frequent words are simply more difficult because readers are
not exposed to them as often and are, therefore, less
familiar with them. An example is the word mentioned above,
on 'connection'. Intermediate level university students tend
not to be familiar with this word. On top of that, the word
is written in its kanji form in this passage, a Chinese
character which is probably not familiar to students.

This has been a discussion of some of the major
categories of difficulty when considering vocabulary. In the
passages in Appendix A, the difficult vocabulary items are
marked in bold face with each one worth one point. As I did

in judging the degree of difficulty of Chinese characters, I

89

am evaluating the degree of difficulty of vocabulary based
upon my own experience as a learner of Japanese.
4.1.20. Cultural Assumptions.

The subject matter of the passages used in this study is
not difficult or technical, so passage content generally does
not cause any trouble for the readers of these passages.
However, with respect to "Aruke Orinpikku", a knowledge of
Japanese culture would help the reader understand the
incident where the woman complained about the use of "fight”.
For Japanese, the use of the word "fight" is not unusual
while walking or running in groups. Such a usage helps to
synchronize the activity and also functions as an
encouragement. It is like being your own cheerleader. In
this case, a cultural knowledge of Japan would discourage a
literal interpretation of this word but would interpret the
use of this word as functioning to convey some sort of
encouragement. In connection with this incident, sentence 13
in this passage indicates that as a result of the woman‘s
criticism Mr. Kaneko "hung his head". This would indicate a
loss of face on his part. A knowledge of this aspect of
Japanese culture would also aid in understanding Mr.Kaneko's
reaction to this event.

With respect to "Hoogen to Kyootsuugo", a cultural
understanding would help the reader understand that Aomori

and Kagoshima refer to names of prefectures in Japan and

90

furthermore that Aomori is in the far north and Kagoshima is
in the far south. Therefore, it is reasonable to expect that
people from these two respective regions would have different
cultural traditions and also would have trouble understanding
each other, given the fact that these two regions are so far
apart, thus making communication between the two difficult in
former times when transportation was an arduous task. Each
name is worth one point of difficulty.
4.2. The Evaluation of Two Texts. In the light of the
twenty features listed above, the two texts, ”Aruke
Orinpikku" ("Walking Olympics") and "Hoogen to KYootsuugo”
(”Dialects and the Common Language”), were evaluated as to
their level of difficulty. Potential grammatical problem
areas in the sentences of these passages were indicated by
the number of the category which was affected. Potential
problem areas with respect to vocabulary, Chinese characters,
rhetoric, and culture were also indicated. After each
sentence was evaluated for problem areas, the total number of
points for that sentence were added up. After all the
sentences of each passage were evaluated, the total
difficulty for each passage was then calculated.

To understand how the passages were analyzed, the
following information is important: the numbers below the
lines equal the number for the category of the grammatical

difficulty (each difficulty = 1 point); bold-face type

91

indicates a vocabulary problem (each problem = 1 point);
underlining indicates a problem with a Chinese character
(each problem = 1 point); and brackets ([...]) indicate a
problem with an idiom (each idiomatic construction = 1
point).

For ease of reference, the following is a repetition of
the list of the grammatical categories used in the analysis

(the same list as in chapter 3, section 3.1):

Sister-
1. Homophonous particles.

a. n1 marking benefit, goal, locative, purpose, result,
the agent, or the indirect object.

b. to marking the conjunctive or the comitative.
c. o marking the direct object or the place traversed.

d. no indicating the locative genitive or the objective
genitive.

e. .mo meaning 'also', 'even‘, or 'as much as'.
2. Nominalizers -- koto and no.
3. Anaphora.
4. Tense.
5. WOrds per sentence.
6. Noun head modification.
7. Conjunctions.
8. Subordination.

9. Passive.

92

10.
11.
12.

13.
14.

15.
16.

Causative.

verb complexes -- verb compounding.

Quotations -- Quoted material comes before the quotation

particle to.

Relative clauses.

Hedges.

a. kamoshirenai 'maybe'.

b. daroo 'perhaps, probably‘.
Further embeddings.

Comparatives.

Difiggflfifiém

17.

Rhetorical structures.

zecabularr,

18.
19.

The percentage of kanji in a passage.

Vocabulary.

a. Formemeaning relationships between English and

Japanese.

1) words similar in form and meaning.

2) WOrds similar in form but different in meaning.

3) WOrds different in form but similar in meaning.

b. Idioms.

Qfllfiflifin

20.

Cultural assumptions.

As an example of how a sentence is analyzed,

sentence 26 from FAruke Orinpikku":

93

consider

(46) Mainen no koto na no de, juumin to sankasha wa
en de musubareru to in.

Since it is an annual event, (they) say that the

residents and participants are strongly tied

together.‘
This sentence would be analyzed as follows:

.Mainen no koto na no de, jun-1n to sankasha wa [g2 de
8
lmgggpareru] to in.

9 12
For this sentence, the numbers underneath indicate three
hypothesized areas of grammatical difficulty involving: 1)
subordination (category 8), 2) the passive (category 9), and
3) a quotation (category 12). Of course, what is difficult
is considered in light of what is difficult for the subjects
of this pilot study. The boldface type for the words jun-1n
and em indicates difficult vocabulary, together contributing
two points of difficulty with respect to vocabulary. Added
to this vocabulary difficulty is the idiomatic phrase [en de
musubareru] ‘to be tied by connection', so there are a total
of three vocabulary problems. in and whareru are
underlined because in the Japanese text (see Appendix B to
see the passage written in Japanese orthography) difficult
Chinese characters are used for these words, so there are two
points of difficulty with respect to Chinese characters. As
can be seen, there are also 15 words in this sentence. Given

this analysis, the results can be tabulated as follows:

94

Grammar = 3 points
vocabulary = 3 points
Kanji = 2 points
Length = 15 words

Refer to Appendix A for a complete analysis of each
sentence in both passages. Under each sentence, there is a
tabulation of the results of the analysis. A discussion of
the results of this analysis is in the next section.

4.3. The Results of the Evaluation.

4.3.1. The Results for "Aruke Orinpikku". The following is
the summary of the data derived from the passages in this
pilot study. The passage "Aruke Orinpikku” has a length of
329 words with an average sentence length of 9.97 words. The
total number of problem points in the passage is 199. These
problem points are broken down as follows:

1) The total number of grammar points is 85 which is
the sum of all the grammatical problems in the
passage. When the total number of grammar points
(85) is divided by the total number of problem
points (199) in the passage, grammatical problems
constitute 42.71% of the possible errors. The index
of grammatical difficulty is the result of dividing
the grammar points (85) by the number of words in
the passage (329). In this case, the index of
grammar difficulty is .258.

2) The total number of vocabulary points is 49 which
is the total of all the vocabulary problems. When
this number is divided by the total problem points,
vocabulary problems are 24.62% of the possible
errors. When the vocabulary points (49) are divided
by the number of words in the passage, then index of
vocabulary difficulty comes to .149.

3) The total number of kanji points comes to 62. When
this number is divided by the total problem points,
kanji problems constitute 31.16% of the possible
errors. The index of kanji difficulty is .188.

95

4) Rhetoric Points only amount to 2 points which comes
to 1.01% of the possible errors. The index of
rhetoric difficulty comes out to .006.

5) There is only 1 culture points which is 0.50% of the
possible errors. The index of culture difficulty
equals .003.

See Chart 1 for a pictorial comparison of the relative
percentages of problems involving grammar, vocabulary, kanji,
rhetoric, and culture. The total passage difficulty for
”Aruke Orinpikku" is calculated by adding the total problem
points (199) to the average sentence length (9.97) and then
dividing that sum by the passage length in words (329), which
results in an index of passage difficulty of 0.64.

Table 1 tabulates the above results. The numbers in
boldface across the top indicate the 20 categories and the
numbers along the sides indicate the sentence numbers. The
row for totals directly below sentence 33 gives the totals
for each category. Given these totals, it is possible to
rank the grammatical categories to determine the order of
their predicted influence on difficulty. The ranking for the
grammatical categories in terms of magnitude of predicted

influence is (excluding categories 5, 17, 18, 19, and 20)

(see page 100):

99329911. NEEDQILQI_EIQDl§m§.
(1) Subordination (8) 16
(2) Relative clauses (13) 15
(3) Quotations (12) 11
(4) Passive (9) 9
(5) NOun head modification (6) 7
(6) Nominalizers (2) 6

96

Chart 1

Percentages of Predicted Problems in 'Aruke
Orinpikku'

Culture
0.50%

   

Kanji 31.16%

Grammar
42.71%

Rhetoric
1.0156

  
  

\

\\
Vocabulary \d/
24.62%

97

Table 1

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

000000100000000000000000000000000 1 89863
0 54.800
2 21.1.00
9 101113212202200111113321132034220 9 a ..
8 101222033211401201412320124335420 Q Wimgmv.
.I .10 t
”4' Cl
..I T11
7 000000000000000000000000010000001 2 mmwcfiw
6 000000000000000000000100000000000 1 dYMIW
cl r a
F8W011
5 000001000000000000000000000000000 1 deHm
cl halo
manltml
4 000000000000000000000100000000010 2 mmmmm
I.- GVKRC
m 3 001001001000300000011011100111010 5
I CI-
]. 020000001010200001000010011100000 19592214
.w n 19846 6.
W Yul 000001000000100000000100000000010 4.. ...
U .ll 8...... ..w
... . ..
ska m0 000000000000100000010000000000000 ZWWWWWW
' I
.l 00. Wre
M. £9 120001000000100010000100011000000 9mmWti
D. ...IaoLt
..l c 8008.1
1' 10' 1|!" 1101' Ila-l] f
N 8 00 00000 00 00 00020 000003 mOWWMWW=
mflltlclf
e 7 000000000100100000000000000000010 3000000
527.227.
6 101001000000110000000000000001100 7
y
114333165857425544678358956768578 9722332“
5 397.6.1.0.S.C
Q4110"
23 ....u
4. 000000000010100000.000010000000000 3. ..D
... S
3 000000010000000000000000000000000 1h$l..l..uo..m L
elm 33
g s .l.
2 000000000010020000000100000000020 6mbmmomom
1.08M]?
.1 .D
I 001000000000001100000100000000000 4 .meOP
1 .p r
u em .“Ntmro.
n 800M]
8 SWWWMWFWX
t 18
n crammed
.M 012345678901234567890123 MW
123456789111111111122222222223333 TAXZSSSI

 

 

 

98

(7) Homophonous particles (1)
(8) Verb complexes (11)

(9) Tense (4)

(10) Conjunctions (7)

(11) Causative (10)

(12) Hedges (14)

(13) Anaphora (3)

(14) Further embeddings (15)
(15) Comparatives (16)

HHI—‘NNUUIfifi

The categories of subordination (8) and relative clauses (13)
are predicted to have the most influence on difficulty since
they predict 16 and 15 problems respectively. Next comes
quotations (12) with 9 problems, and so on. What this
supports is that for this passage, subordination is a more
important predictor of difficulty than conjunctions. These
are predictions that need to be tested empirically, which
will be done at a later stage in this pilot study.

4.3.2. The Results for ”Hoogen to Kyootsuugo". The
following is the summary of the data derived from an analysis
of "Hoogen to Kyootsuugo". The total Passage Length is 234
words with the average sentence length of 21.27 words. The
total number of problem points in the passage is 88. These
problem points are broken down as follows:

1) The total number of grammar points is 54 which is
the sum of all the grammatical problems in the
passage. When the total number of grammar points
(54) is divided by the total number of problem
points (88) in the passage, grammatical problems
constitute 61.36% of the possible errors. The index
of grammatical difficulty is the result of dividing
the grammar points (54) by the number of words in
the passage (234). In this case, the index of
grammar difficulty is .231.

2) The total number of vocabulary points is 22 which
is the total of all the vocabulary problems. When

99

this number is divided by the total problem points,
vocabulary problems are 25.00% of the possible
errors. When the vocabulary points (22) are divided
by the number of words in the passage, then index of
vocabulary difficulty comes to .094.

3) The total number of kanji points comes to 8. When
this number is divided by the total problem.points,
kanji problems constitute 9.09% of the possible
errors. The index of kanji difficulty is .034.

4) Rhetoric Points only amount to 2 points which comes
to 2.27% of the possible errors. The index of
rhetoric difficulty comes out to .009.

5) There is only 2 culture points which is 2.27% of
the possible errors. The index of culture
difficulty equals .009.

See Chart 2 for a pictorial comparison of the relative
percentages of problems involving grammar, vocabulary, kanji,
rhetoric, and culture. The total passage difficulty is
calculated by adding the total problem points (88) to the
average sentence length (21.27) and then dividing that sum by
the passage length in words (234), which results in an index
of passage difficulty of 0.47.

Table 2 tabulates these results. As with Table 1, the
numbers in boldface across the top indicate the 20 categories
and the numbers along the sides indicate the sentence
numbers. The row for totals directly below sentence 11 gives
the totals for each category. Given these totals, it is
possible to rank the grammatical categories to determine the
order of their predicted influence on difficulty. The

ranking for the grammatical categories in terms of magnitude

of predicted influence is (excluding categories 5, 17, 18,

100

Chart 2

Percentages of Predicted Problems in 'Hoogen
to Kyootsuugo‘

  

Culture
Kanji 9.09% 2-2
Rhetoric
2.27%
/
Vocabulary
25.00%

Grammar
61.36%

101

Table 2

Problems

01

8910112131415161181920

0

00000

9
10
11

0—000
oooooo-O-
mm—do-Moo-
-o-—oooo-—-o
mooo~o~oo
ooooo—o-oo
ooooooooo
oooooo-oooo
uo~m~oo~o~
00000000000
00000000000
doooo~ooooo
Noooooroooso
NJ—OOO-‘(N--0‘N

otels 4 4

N
O

410

A
O
N

ver sentence . Sum of
of . obiems a . of . Grammar difficul .
voc. a . of voc. Vocabu cllf icul

2 of ken i lems = . of kan 1 Ken 1 difficu

2 of rhet. problems = 2 of rhet. prob. = Rhetorical difficulty .009

 

of cu lure rob s = of cu ture ltural 01 ficul =

Index of Difficu

102

19, and 20):
QéEQQQIX. 0 0b

(1) Relative clauses (13) 11
(2) NOun head modification (6) 10
(3) Subordination (8) 10
(4) Homophonous particles (1)
(5) Nominalizers (2)
( 6) Conjunctions ( 7 )
(7) Passive (9)

(8) Quotations (12)
(9) Tense (4)

(10) Anaphora (3)

(11) verb complexes (11)
(12) Causative (10)

(13) Hedges (14)

(14) Further embeddings (15)
(15) Comparatives (16)

OOOOHHNwfihfiF

The category of relative clauses (13) is predicted to have
the most influence on difficulty since it predicts 11
problems. Next come noun head modification (6) with 10
problems and subordination (8) with 10 problems. Then comes
homophonous particles (1) with 4 problems and so on. This
says that for this passage, relative clauses are a more
important predictor of difficulty than homophonous particles.
These are predictions for this passage that need to be tested
empirically at a later stage in this pilot study.

4.4. Summary. According to the features given above, the
jpassage "Aruke Orinpikku" is more difficult than "Boogen to
Kyootsuugo”. This can be seen from the following table of

difficulty indices:

103

Table 3

WW

Grammar .258 .231
VOcabulary .149 .094
Kanji .188 .034
Rhetoric .006 .009
Culture .003 .009
Passage Difficulty .64 .47

In the areas of grammar, vocabulary, and kanji, the indices
of difficulty predict that "Aruke Orinpikku" will be more
difficult than "Boogen to Kyootsuugo". Particularly, there
is a large difference in the indices of kanji and vocabulary
difficulty in these two passages. This would predict that a
large measure of the difficulty between these two passages is
predictable from the differential in kanji and vocabulary
difficulty. Thus, passage difficulty is largely a function
of kanji and vocabulary difficulty in the case of these two
passages. However, other factors, such as linguistic
complexity, should not be ignored since they, too, play a
role in textual difficulty. Also, as explained above,
average sentence length is incorporated into the calculation
of passage difficulty. It is interesting to note that the
average sentence length for ”Aruke Orinpikku" (9.97) is

shorter than that for "Hoogen to Kyootsuugo" (21.27).

104

However, ”Aruke Orinpikku” is rated as more difficult by the
index of passage difficulty. Also the index of grammar
difficulty as given above would rate "Aruke Orinpikku" (.258)
as more difficult than "Hoogen to Kyootsuugo" (.231). This
would indicate that sentence length by itself is not
necessarily a reliable predictor of passage difficulty. An
empirical study needs to address these observations.

The above data also make certain predictions about the
relative contribution of grammar, vocabulary, .kanjiq
rhetoric, and culture percentage-wise to the total number of
possible problem areas. The following table shows the
percentages each area contributes to the total errors for
each passage:

Table 4

WW

Grammar 42.71% 61.36%
VOcabulary 24.62% 25.00%
Kanji 31.16% 9.09%
Rhetoric 1.01% 2.27%
Culture 0.50% 2.27%

The following chapter discusses the results of an initial
experimental pilot study to examine some of the above

predictions. This experiment is preliminary to a computer

105

experiment which attempts to further investigate these

issues.

106

CHAPTER 5

The Experiment To Assess Difficulty in Japanese Text

5.0. Introduction. The two passages, "Aruke Orinpikku"
("Walking Olympics") and "Hoogen to Kyootsuugo" ("Dialects
and the Common Language"), will now be investigated as to
their actual difficulty. In this pilot study, an experiment
was conducted wherein the subjects circled or underlined the
segments in the passage that they thought were difficult.
This information was evaluated to give a preliminary
assessment of the predictions that were made in chapter 4 (a
further assessment of these predictions will be given as a
result of the computer experiment in chapter 6).
Furthermore, this experimental technique was critiqued as to
its validity in assessing textual difficulty. The question
of whether this experimental instrument measures what it is
supposed to is discussed in this chapter. The purpose of
this investigation is to facilitate the creation of reading
software based on these two passages. In order to be
effective, software must help students where they are
actually having reading difficulty. This investigation was
conducted to find these areas of actual difficulty, and thus
help streamline the software design process. This software,

along with teaching reading comprehension, will be able to

107

further evaluate reading difficulty in the process, thus
providing a feedback mechanism for the evaluation of textual
difficulty

The subjects in this pilot study were six third-year
Japanese students at a large Midwestern University. They
were told to circle or underline segments in these passages
that caused difficulties in the five areas of grammar, kanji,
vocabulary, rhetoric, or culture. They were asked not to
distinguish between these areas of difficulty by the way they
marked the segments because the distinction between these
five areas was probably not clear in their minds. The
contribution of these areas was evaluated by an interpretive
analysis of the way the students marked the text. The
students were not asked to mark the passages within a certain
time period; however, they were finished within 25 minutes.

The passages themselves were presented in a typical
Japanese format in vertical columns and no spacing between
words. The characters were presented in large enough
characters for the students to easily read and mark. Samples
of the texts used for the experiment are presented in
Appendix B.

The following is a discussion of how the experiment
meets the predictions of the previous chapter. The original
analysis is presented along with the experimental results for

each sentence. The fact that this experiment did not fulfill

108

certain predictions does not imply that the predictions were
bad. Perhaps, the experimental instrument for measuring them
is inaccurate or invalid in certain ways. This issue is
discussed further in this chapter along with a proposal for a
better experimental instrument. The following is an analysis
of experimental results for "Aruke Orinpikku” and "Hoogen to
Kyootsuugo". The analysis is based on two calculations:

1) The first calculation is the percentage of
predictions made in the previous chapter that were
correct. This is called the prediction success
rate.

2) The second calculation is the percentage of the
ag§g§;_problem areas as determined by this
experiment that were successfully predicted. This
is called the actual success rate.

By looking at these two types of success rates in
combination, it is possible to acquire a comprehensive view
of the accuracy of the predictions. The following passages,
FAruke Orinpikku” and "Hoogen to Kyootsuugo", were analyzed
using these two types of success rates. It should also be
noted that the numerical results of the student survey are
printed in outline characters immediately above and below the
sentences in section 5.1. The numbers printed in outline
above the line indicate the number of students who had
trouble with the particular Chinese character in question.
The numbers in outline below the line indicate the number of

students who had either problems with vocabulary or grammar.

Sometimes, the distinction between grammar and vocabulary

109

problems is not so clear due to the fact that they are marked
the same way in this experimental method. It is a matter of
interpretation. The lowest line of numbers below a line of
text which is printed in ordinary characters contains the
predictions made in the previous chapter. Whole phrases
which students circle must be treated as a whole. To
indicate this grouping, they are underlined. In the
following, the use of the Japanese word kanji refers to
Chinese characters. The word kanji itself has the following
literal meaning: kan 'Han (the old name for China)’ plus ji
'letter'. Each sentence is accompanied by two tables: 1)
Predictions and 2) Actual Problems. The Table of Predictions
breaks down the predictions into their respective categories:
grammar, vocabulary, kanji, rhetoric, and culture. The
number of fulfilled predictions in each category is listed
along with the prediction success rate. Finally, the total
number of predictions and fulfilled predictions is added up
and the total prediction success rate for the whole sentence
is calculated. The Table of Actual Problems breaks down the
actual problems into their respective categories as in the
Table of Predictions. The number of actual problems in each
category that were successfully predicted is listed along
with the actual success rate. Finally, the total number of
actual problems and the total number of the actual problems

that were successfully predicted are respectively added up.

110

Then, the total actual success rate is calculated by dividing
the total number of actual problems that were successfully
predicted by the total number of actual problems. Following
the analysis of all the sentences in the passages, there is a
general discussion of the results for these passages, after
which some general conclusions produced by the analysis are
stated.

5.1. Data Presentation.

5.1.1. The Analysis of "Aruke Orinpikku". The following is

a sentence-by-sentence analysis:

4
l. Mainen oranda no chiisa na machi de ooki na ggzggshi ga
2
6
okonawareru.
l
9

'In a small town in Holland, a big event is held every
year.‘

Four of the six students circled the kanji, moyoo
'event', to indicate it as a source of trouble; two of the
six students marked the whole word, moyooshi. The choice of
'the whole word also appears to indicate that the students had
'tronble reading and understanding the kanji. This indicates
'that.both moyoo and moyooshi are actual problem areas. One
student also marked okonawareru 'is held' as a source of

difficulty which could mean that he or she failed to

ill

understand the passive morphology (grammatical feature 9).
Thus okonawareru is the third actual problem area. The way
the students marked the sentence indicates that the
predictions were fairly accurate. The experiment confirmed 3
of the 4 predictions at a prediction success rate of 75%.
The only unfulfilled prediction involved noun head
modification (grammatical feature 6). The actual success
rate was 100% since all of the three actual problems were
predicted. The results for this sentence are tabulated in

the following table:

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 2 1 50%
VOcabulary 1 1 100%
Kanji 1 l 100%
Total 4 3 75%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar l 1 100%
Vocabulary l 1 100%
Kan ji 1 1 100%
Total 3 3 100%

112

2. FoodeezulMaachi to mo yobare, aruke orinpikku to mo

1 l
12 9 12
yobarete iru.
l
9

'It is called the four days' march; it is also called the
walking 01ympics.’

In this sentence, Foodeezu 'four days',.Maachi 'march'
and yobarete 'is called' were marked only once indicating
three actual problem areas. The two-word phrase, Foodeezu
Maachi, was not predicted to cause trouble since it is
directly transliterated from English. The fact that yobarete
was marked seems to indicate a grammatical problem with the
passive, so one grammatical prediction (grammatical feature
9) was successful. Only 1 out of the 4 predicted grammatical
problems was verified; therefore, the prediction success rate
for this sentence was 25%. Also, only one actual problem
area was predicted, so the actual success rate was 33% since

there were three actual problems. (W,

abbreviated as 31;, means no predictions were made in chapter

4, so no fulfillments are possible.)

113

Predictions

Prediction
Success
Rate

25%
not applicable
n/a

25%

Actual
Success
Rate

100%
0%
n/a

33.3%

Fulfilled
Predictions Predictions
Grammar 4 1
Vocabulary 0 0
Kanji 0 0
Total 4 1
Actual Problems
Actual
Actual Problems
Problems Predicted
Grammar 1 1
VOcabulary 2 0
Kanji 0 0
Total 3 l
3 3
3. Nimannin o figpru sekai kakkoku no hitobito ga kono machi
2 1 l
1 13 6
l

Neimeehen ni atsumatte, aruku.

l

1

'More than 20,000 people from various countries throughout
the world gather in the town of Nijmegen and walk.'

Three students marked the kanji in the word, koeru

'exceed', and two marked the whole word indicating a possible

problem with meaning.

This indicates two actual areas of

difficulty, one for the kanji and one for the whole word.

One student marked sekai 'world', a word which was not

114

predicted but apparently is the third actual problem. Also
three students marked the kanji, kaklu) 'each, various', in
kakkoku 'various countries' (the fourth actual problem), and
one student marked the whole word (the fifth actual problem).
Trouble with this word was not predicted. One student marked
Neimeehen 'Nijmegen' (the sixth actual problem), since he or
she did not realize it is the name of the town where the walk
was held. One student marked the kanji in the word atsumatte
'gather' (the seventh actual problem), and another marked the
whole word (the eighth), perhaps indicating trouble with the
embedded verb (grammatical feature number 8: subordination).
Only 3 of the 6 points of difficulty predicted in chapter 4
were verified by this experiment (see the table below). The
prediction success rate was 50%. In addition, the experiment
indicated that were were five problem areas that were not
predicted, which meant an actual success rate of 37.5% (see

the table below).

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 4 1 25%
VOcabulary 1 1 100%
Kanji 1 1 100%
Total 6 3 50%

115

Actual Problems

Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 1 1 100%
VOcabulary 4 1 25%
Kanji 3 1 33.3%
Total 8 3 37.5%

4- mm mo nai-
5
'There are no winners or losers.‘

Five students marked shoohai 'victory or defeat' as
causing trouble which indicates that the meaning of the word
is a comprehension problem for the students. Also, the
kanji, shoo 'victory', and the kanji, hai 'defeat', could be

a problem, but that is not indicated here.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate

Grammar 0 0 n/a
VOcabulary 1 1 100%
Kanji 2 O 0%
Total 3 1 33.3%

116

Actual Problems

Actual Actual
Actual Problems Success

Problems Predicted Rate
Grammar 0 0 n/a
Vecabulary 1 1 100%
Kanji 0 0 n/a
Total 1 1 100%

4 3
5. M mo nai.
1

'There is no order of placement.’

Four students marked the kanji, jun 'order', as an area of
difficulty, and three students marked the lkanjiq :i
'position'. In addition, one marked the whole word, jun'i
'order', indicating a problem in understanding this word.
Thus, there were three actual problems. The prediction

success rate and the actual success rate were 100%.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 0 0 n/a
VOcabulary l 1 100%
Kanji 2 2 100%
Total 3 3 100%

117

Actual Problems

Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
VOcabulary 1 1 100%
Kanji 2 2 100%
Total 3 3 100%
l
6. YOkkakan arukitooshita mono ni kappa no ggggsho to
3 3
11 13 6
2
medaru ga ataerareru dake da.
2
9 15

'It's just that a certificate of completion and a medal
are given to those who last the distance over four days.’

No students marked arukitooshita 'continued to walk'
although it was predicted to be a problem. Three students
marked kanpo 'completion of walking', perhaps indicating a
problem with the kanji. However, it cannot be stated that
there was an actual problem with the kanji since the students
marked the whole word and the individual Chinese characters
in this word are not that difficult for intermediate
students. All that can be said is that the students had
'trouble with the vocabulary item, kanpo. One student marked
the kanji, shoo 'proof', and three students marked the

cxmmplete word, shoosho 'certificate'. Two marked the kanji

118

in the word, ataerareru 'be given' (not predicted to cause
trouble in chapter 4); furthermore, two circled the whole
word, perhaps indicating trouble with the passive
construction (grammatical feature 9). The total number of

actual problems was 5.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 5 1 20%
Vecabulary 3 2 66.7%
Kanji 2 l 50%
Total 10 4 40%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 1 1 100%
VOcabulary 2 2 100%
Kanji 2 1 50%
Total 5 4 80%

7. KOtoshi no taikai ni wa nihon kara gojuurokunin sanka
4

shite.
'Fifty-six Japanese participated in this year's march.’
Four people marked the word sanka 'participation' as a
source for a comprehension problem. That is all that was

marked. The number of actual problems was 1 .

119

Predictions

 

Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar O 0 n/a
Vocabulary 2 1 50%
Kanji 0 0 n/a
Total 2 1 50%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
vocabulary 1 1 100%
Kanji 0 0 n/a
Total 1 1 100%
3
8. Sanka kyuukai no Kaneko mgmgkazg,san (Nihon.Aruke.Aruke
3 l 2 l
——L

1
fizggkaichoo) ni we konna omoide ga aru.

 

 

_2;__ 3

'Tomokazu Kaneko, chairman of the Japan walk-walk Society,
who has participated nine times, has the following
recollection.
Three people marked sanka 'participation', one person
kyuukai 'nine times' and two people Kaneko, which is a family

name. All these problem areas were not predicted. Three

people marked the kanji, Temo (a Chinese character used in a

120

personal name), as troublesome, which was predicted. One
person marked the whole phrase, Nihon Aruke Aruke Kyookaichoo
'chairman of the Japan Walk-Walk Society', and one marked
iAruketAruke.Kyookaichoo 'chairman of the Walk—Walk Society'.
In addition, Kyookaichoo 'head of the society' was marked by
two people, clearly a vocabulary problem involving difficult
kanji. The rest of the phrase, Nihon Aruke Aruke 'Japan
Walk-Walk', contained vocabulary and kanji that was simple
for the subjects, so perhaps it was a problem with
concatenation. One person had a problem with the kanji, de
'appearance', in omoide 'recollection'. The total number of

actual problems is 6.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 1 0 0%
VOcabulary 1 1 100%
Kanji 3 1 33.3%
Total 5 2 40%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
Vocabulary 4 1 25%
Kanji 2 1 50%
Total 6 2 33.3%

121

2 l
9. Hajimete sanka shita toki, nakama to issho ni "faito,

3 3 4
13
1 3
faito" to gogggggg ni gaggggg aruita.
l l
12 8

'The first time he participated, he shouted with other
participating Japanese, "Fight! Fight!"'

Three people marked sanka 'participation'. Two people
had trouble with naka 'relationship' in nakama 'companion'
while three people marked the whole word. One person did not
understand the character, sho 'thread', in issho 'together
with' while three people marked the whole word. One person
marked kowa 'voice' in kowadaka 'loud voice', and one person
marked the whole word. Probably, more than one person should
have marked kowa since it is a very difficult reading;
however, the meaning of this kanji, 'voice', is not
difficult, so the students assumed they knew how to read it.
Another word was difficult because of the kanji contained in
it: sakende 'shout'. Three people marked the kanji in this
word, and one person marked the whole word because he or she
perhaps had trouble with the embedding signalled by the
gerundive "sakende", as explained in the previous chapter
(grammatical feature 8). Thus, the total number of actual

problems is 9.

122

Grammar
Vocabulary
Kanji

Total

Grammar
Vocabulary
Kanji

Total

7

'Then an old woman came forward and said:'

Fulfilled
Predictions Predictions
3 l
2 2
3 2
8 5

Actual
Problems

to hubhd

4

Predictions

Actual Problems

Actual

Problems
Predicted

U1 NNl—i

1 2
10. Suru to, nonpgi no josei ga suggmidete itta.

Prediction
Success
Rate

33.3%
100%
66.7%

62.5%

Actual
Success
Rate

100%
80%
100%

55.6%

Four people marked nenpai 'elderly', one person marked the

kanji, sei 'gender', in josei 'woman', and two people marked

the kanji in susumidete 'come forward', so the total number

of actual problems is 3.

123

Grammar
Vocabulary
Kanji

Total

Grammar
VOcabulary
Kanji

Total

Predictions

2
2
2
6

Actual

Problems

0
1
2
3

Predictions

Fulfilled
Predictions

O
1
1

2

Actual Problems
Actual

Problems
Predicted

0
1
1
2

11. "Naze tatakaou to in no."

4 12

___._1.___

2

"Why do you talk about fighting."

Four students marked the kanji in tatakaou

Prediction
Success
Rate

0%
50%
50%

33.3%

Actual
Success
Rate

n/a
100%
100%

66.7%

fight'. One student marked the phrase, tatakaou to in,

"Let's fight."', which indicates that the student had trouble
with the grammar with respect to the verb tense of tatakaou
(grammatical feature 4) and the structure signalled by the

quotative particle , to,

(grammatical feature 12).

tOtal number of actual problems is 3.

124

'let's

Thus the

Grammar
Vocabulary
Kanji

Total

Grammar
Vocabulary
Kanji

Total

Predictions

3
0
1

Actual
Problems

2
0
1

3

Predictions

Fulfilled
Predictions

2
0
1

3

Actual Problems

Actual
Problems
Predicted

2
0
1

3

Prediction
Success
Rate

66.7%
n/a
100%

75%

Actual
Success
Rate

100%
n/a
100%

100%

4

12. "watakushitachi wa [ggggkgi wa gonen] da we."

"we don't want to fight."

Four students marked the kanji in tatakai 'fight',

the total number of actual problems is 1.

Grammar
ikxcabulary
Kazyfli

fmytal

Predictions
Fulfilled
Predictions Predictions
O 0
1 O
1 1
2 1

125

_.1,,_

Prediction
Success
Rate

n/a
0%
100%

50%

SO

Actual Problems

 

Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
VOcabulary 0 0 n/a
Kanji l 1 100%
Total 1 1 100%
3
13. Ganbarou to iikikaseru karui chooshi no "faito" datta
1
4 12 11 10 13 6
1
ga, Kaneko san we "we don't want to fight" to in kotoba ni
7 12 13
4 1
honorareta mono o sagshite, fukaku atama o sageta.
1 1 1 3
9 l3 8

'It was a light "Fight!" calling on people to exert an
effort, but Kaneko realized the thinking behind the words,
"we don't want to fight," and hung his head.‘

One student marked ganbarou 'let's try hard', indicating a
problem with the verb tense (grammatical feature 4). One
student circled iikikaseru 'persuade', perhaps indicating at
least problems with the grammatical form. The grammatical
form of this word can provide potentially three problem

areas. First of all, iikikaseru is a verbal complex

(grammatical feature 11):

126

ii + kikaseru
say make to hear.

Secondly, kikaseru is the causative form (grammatical feature
10) of the verb kiku 'hear'. Thirdly, iikikaseru is a verb
in a relative clause (grammar feature 13). As for explaining
the difficulty involved with iikikaseru for this particular
student, it is possible to hypothesize exactly what the cause
is. All that can be said is that there are three
possibilities for the grammatical source of the trouble: 1)
not knowing the structure of the verbal complex, 2) not
understanding the causative form, or 3) not understanding it
is the form of the verb used in a relative clause (not a very
likely explanation since these are third-year students). Any
one of these grammatical causes could explain the difficulty
with iikikaseru, so any one of these predictions can be
fulfilled. However, this experimental tool cannot determine
which prediction is the explanatory cause for the difficulty
with this word. To continue with the analysis of this
sentence, three students marked the kanji in karui 'light
(adj.)'. One student marked the word chooshi 'style,
manner', perhaps indicating a problem with vocabulary. One
student circled the complete phrase Ganbarou to iikikaseru
karui chooshi. It seems like this indicated a grammatical
problem with the head-modifier construction for the student.
The complete construction is really Ganbarou to iikikaseru

karui chooshi no "faito" 'It was a light ”Fight!” calling on

127

people to exert an effort'. The difficulty in this
construction was predicted by the grammatical features listed
for the items in this sentence (see sentence 13 above): 1)
the tense of ganbarou (grammatical feature 4), 2) the
quotation particle to (grammatical feature 12), 3) the verb
complex iikikaseru (grammatical feature 11), 4) the causative
form kikaseru (grammatical feature 10), and 5) the embedding
caused by the relative clause ganbarou to iikikaseru
'persuade to exert an effort' (grammar feature 13). Later in
the sentence, one student circled the kanji, ha 'leaf‘, in
kotoba 'words' ('h' changes to 'b' because of sequential
voicing). One student marked komerareta 'be included',
perhaps due to grammatical difficulties caused by the passive
(grammatical feature 9). Four students circled the kanji in
sasshite 'realize', and one student circled the whole word
perhaps indicating trouble with verb form because it
indicates an embedded clause (grammar feature 8). One
student marked the kanji in fukaku 'deeply' as causing
trouble. Another student marked the whole word, fukaku,
again perhaps indicating trouble with the kanji since in the
text the entire word is composed of the kanji plus the
adverbial suffix, which is relatively easy. Finally, three
students marked sageta 'lowered', indicating problems with

vocabulary. Thus, the total number of actual problems is 13.

128

This experimental instrument picked

interference in this passage.

Predictions
Fulfilled
Predictions Predictions
Grammar 12 7
VOcabulary 2 1
Kanji 4 2
Culture 1 0
Total 19 10
Actual Problems .
Actual
Actual Problems
Problems Predicted
Grammar 7 6
Vocabulary 2 1
Kanji 4 2
Total 13 9

up no cultural

Prediction
Success
Rate

58.3%
50%
50%

0%

52.6%

Actual
Success
Rate

85.7%
50%
50%

69.2%

14. Aruku koto no daisuki na hitobito ga atsumatte, aruku

2 6
koto o tanoshimu.

2

8

'People who really like walking, gather and enjoy their

‘walking.'

In this passage no actual problems were indicated.

‘Therefore, no predictions were fulfilled.

129

Predictions

Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 4 0 0%
VOcabulary 0 0 n/a
Kanji 0 0 n/a
Total 4 0 0%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
VOcabulary 0 0 n/a
Kanji 0 0 n/a
Total 0 0 n/a
1
15. fligggi no aka 0 aruku.
__.1___

1
'They walked over green hills.‘

For one student, the kanji for oka 'hills' was a
problem. Four students circled midori no oka 'green hills'
indicating difficulty in understanding this phrase. It seems
that the subjects had trouble understanding the Chinese
characters for midori 'green' and oka 'hills' because as
vocabulary items they are quite simple. This means the
number of actual problems is two: one involving the Chinese

character oka and the other the character midori.

130

Predictions

Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 1 0 0%
vocabulary 0 0 n/a
Kanji 1 1 100%
Total 2 1 50%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
Vocabulary 0 0 n/a
Kanji 2 1 50%
Total 2 1 50%

16. gaggsoi no michi o aruku.

1

1

'They walked on roads alongside canals.‘

All the subjects had trouble with ungazoi 'alongside
canals'. Four students circled unga 'canals'. It seems that
the students had trouble in understanding the kanji, un
'transportation', and the kanji, ka 'river' (which becomes ga
through sequential voicing). One student circled 201'
"along', which is a noun derived from sou 'to go along' (/s/

changes to /z/ because of sequential voicing). Because 201

131

is not written in kanji, it is a vocabulary problem. The

result is that there are three actual problems.

Predictions
Fulfilled
Predictions Predictions
Grammar 1 O
vocabulary 1 l
Kanji 2 2
Total 4 3
Actual Problems
Actual
Actual Problems
Problems Predicted
Grammar 0 0
Vocabulary 1 1
Kanji 2 2
Total 3 3

17. Arukinagara fureai ga umareru.
2 1
8 9

'While walking, contacts are made.‘

Prediction
Success
Rate

0%
100%
100%

75%

Actual
Success
Rate

n/a
100%
100%

100%

Some subjects marked the separate morphemes, fare

'contact' and ai 'mutual', that compose fureai 'contacts',

(but actually these morphemes form one word.

'they did not understand.

132

It is this word

Predictions

 

Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 2 0 0%
Vocabulary 1 1 100%
Kanji 0 0 n/a
Total 3 1 33.3%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
VOcabulary 1 l 100%
Kanji 0 0 n/a
Total 2 1 50%
1 1 3
18. "Kiiroi ribon," "Oosuzanna," "Moshi, moshi m yo” to
A
1
12
1
samazama na uta ga kikoete kuru.
L __.1.__

 

8

'Various songs, such as ”Tie a Yellow Ribbon," "Oh,
Suzanna," and ”Moshi, moshi Kame yo”, can be heard.‘

One student circled the kanji, ki 'yellow', and another

irr>'color'. Two students did not understand the title for

the Japanese folk song. One circled kame yo 'turtle

(emphasis) ', and the other the complete title, ”Moshi, moshi

133

kame yo" 'Hello, turtle (emphasis)'. It seems like the kanji
for kame 'turtle' was difficult. Perhaps, there also were
difficulties in understanding the other vocabulary in the
title, e.g., moshi moshi 'hello'. To continue with the
analysis of the sentence, one student circled samazama na
'various', which appears to be a vocabulary problem. The
kanji, uta 'song', was circled once. Finally, one student
circled the phrase kikoete kuru, which is a sequence of
verbs, kikoete 'can hear' (gerundive form) and kuru 'come'.
This construction exhibits subordination on the part of
kikoete (grammatical feature 8). The implication of the
verb, kuru, is that the sound is coming toward the narrator.

In summary, the total number of actual problems is 8.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 2 1 50%
vecabulary 1 0 0%
Kanji 1 1 100%
Total 4 2 50%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 1 1 100%
VOcabulary 3 0 0%
Kanji 4 1 25%
Total 8 2 25%

134

1
19. Quake 0 utainagara Qgigai ga aruku.
3 5

8
'Soldiers walk while singing military songs.’

Three subjects circled gunka 'military song', one
subject circled the kanji in utainagara 'while singing', and
five subjects circled heitai 'soldier'. It seems that the
difficulty of the Chinese characters in these words is the
source of the problems with understanding. This difficulty
would make understanding the meanings of these words a
problem. Thus, there are five actual problems here, all

related to difficulties with Chinese characters.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 1 0 0%
VOcabulary l 0 0%
Kanji 4 4 100%
Total 6 4 66.7%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
VOcabulary 0 0 n/a
Kanji 5 4 80%
Total 5 4 80%

135

1 3 2
20. Kuroi hada o kgggzgkaseta musumesan ga aruku.
1 1

A 2.:

10 13

 

'Girls with glistening tanned skins walk.'

One student circled the kanji in kuroi 'black'. Two
students circled the word, hada 'skin', indicating a
vocabulary problem. One student marked da (in hada) which
clearly indicates a confusion with word boundaries. Three
students indicated trouble with the kanji in kagayakaseta
'made to shine'. One student circled kaseta indicating
trouble with the causative suffix ase. This would be a
grammatical problem (grammatical feature 10). Two students
circled the whole word, kagayakaseta, perhaps indicating at
least a vocabulary problem. Two students circled musume
'girl', which seems to indicate a problem in reading the

kanji for 'girl'. Thus, there are a total of seven actual

problems.
Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate

Grammar 2 1 50%
vecabulary 1 1 100%
Kanji 1 1 100%
Total 4 3 75%

136

Actual Problems

Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 1 1 100%
Vocabulary 2 1 50%
Kanji 3 1 33.3%
Total 7 3 42.9%
4 3
21. Ryukku o [§g_ni shita] goojin ga aruku.
1 2
13

'An old man with a knapsack on his back walks.‘

One subject circled ryukku 'rucksack', which was not
written in kanji; therefore, this is a vocabulary problem.
Four subjects circled se 'back'. This kanji was predicted to
be difficult for them. Three subjects circled the kanji, roo
'old', in roojin 'old man'. Two circled the whole word,
perhaps indicating a vocabulary problem. Consequently, the

number of actual problems is 4.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 1 0 0%
VOcabulary 3 2 66.7%
Kanji 2 2 100%
Total 6 4 66.7%

137

Actual Problems

Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
VOcabulary 2 2 100%
Kanji 2 2 100%
Total 4 4 100%
3
22. Orinpikku no ggiggig,wa hank; yori mo kochira ni
2 2
16 1

uketsugarete iru no kamoshirenai.

 

1
11 9 2 14
'The real spirit of the Olympic Games may have been
preserved more in this (four days' march) than in the
Olympics itself.‘

Three students circled the kanji, sei 'feeling', in
seishin 'spirit' signifying a kanji problem while two
students indicated a vocabulary problem by circling the whole
‘word. Two students circled honke 'originator' indicating a
problem with word meaning. One student circled uketsu and
rate iru in the verb phrase, uketsugarete .iru 'has been
preserved, inherited'. There seem to be a number of factors
involved in this mistaken parsing. In effect, this student

parses this verb phrase as follows:

uketsu ga rete iru.

138

The ga in this phrase is treated like the nominative
particle, ga. It seems there is a basic misunderstanding of
the passive structure in this phrase (grammatical feature 9).
Another problem here is that uketsugarete is derived from
uketsugu 'inherit', a verb compound (feature 11), which is
analyzed as follows:

uke + tsugu
'receive' 'succeed'.

All these problems with parsing resulted in putting word
boundaries in the wrong place. This analysis indicates a

total of 5 actual problem areas.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 6 2 33.3%
VOcabulary 3 2 ' 66.7%
Kanji 3 1 33.3%
Total 12 5 41.7%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 2 2 100%
‘v0cabu1ary 2 2 100%
.Kanji 1 l 100%
Total 5 5 100%

139

2 1 2
23. indoo ni wa, isu o narabe, yasunde ike to tangggki o
4 1 2 2 2

8 11 4 11

2
suru shufu ga iru.
1

13

'Along the road, there are housewives who line up chairs
and beckon to (walkers) to come and rest.‘

Two students marked the kanji, en 'along', signifying a
kanji problem, and four marked the whole word, endoo 'along
the road' , to indicate a vocabulary problem. One student
marked isu 'chair', indicating a vocabulary problem since
this word is not written in kanji in the text. One student
marked the kanji in narabe 'line up', and two marked the
whole word indicating difficulty with vocabulary. Two
students marked ike 'go', which seems to indicate a problem
with tense (grammatical feature 4). Two students circled the
kanji for mane 'invite' in the word, temaneki 'beckon', and
two circled the whole word. Finally, two students circled
the kanji for fu 'woman', and one circled the whole word,
shufu 'housewife'. This indicates a total of 10 actual

problems.

140

Grammar
vocabulary
Kanji

Total

Grammar
Vocabulary
Kanji

Total

Predictions

Predictions

5
2
2
9

Actual

Actual

Problems

#wflhi

10

Fulfilled

Predictions

l
2
2
5

Problems

Actual

Problems
Predicted

l
2
2
5

24. Kaohii ya kukkii o furumau ie mo aru.

Prediction
Success
Rate

20%
100%
100%

55.6%

Actual
Success
Rate

100%
40%
50%

50%

'There are families which treat (walkers) to coffee and

cookies.‘

Five students marked the word furumau

' treat ' .

It seems

to indicate a problem with vocabulary, so there is one actual

problem.here.

(arammar
\kxaabulary
imanji

{natal

Predictions
Fulfilled
Predictions Predictions
1 0
l 1
0 0
2 1

141

Prediction
Success
Rate

0%
100%
n/a

50%

Actual Problems

Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
vecabulary 1 1 100%
Kanji 0 0 n/a
Total 1 1 100%
2
25. Kuroobaa no haggtgbg_o okutte kureru shoojo mo iru.
3 1
8 13

'There are girls who present (the walkers) with clover
bouquets.‘

Two subjects circled the kanji, taba 'bunch', indicating
they did not know it, while 3 subjects circled the whole
word, hanataba, 'bouquet' indicating they did not understand
the meaning of the word. Finally, one student circled shoojo
'girls' indicating a lack of understanding of the word's

meaning. As a result, there are three actual problems.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 2 0 0%
Vecabulary 1 1 100%
.Kanji 1 1 100%
Total 4 2 50%

142

Actual Problems

 

Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
vecabulary 2 1 50%
Kanji 1 1 100%
Total 3 2 66.7%
5
26. Msinen no koto na no de, jun-1n to sankasha wa Lg; de
4 3
...—1m _=1
8
2
mbareru] to in .
2
9 12

'Since it is an annual event, (they) say that the
residents and participants are strongly tied together.‘

Four students marked juumin 'residents', and three
marked sanka 'participation' as difficult vocabulary. Note
the sanka was also marked in sentences 7 and 9, so sanka was
a consistently difficult vocabulary item. One student even
marked the phrase juumin to sanka, possibly indicating a
problem with complex noun head forms (grammar feature 4)
because to 'and' connects both juumin and sanka. This
grammatical problem was not predicted. To continue with the
analysis of this sentence, five students circled en

'connection' as difficult, which was predicted. Two students

143

marked the kanji in musubareru 'be tied', while two students
marked the whole word. The fact that the whole word was
marked can lead to the conclusion that the students had
trouble because this verb is in the passive form (grammar
feature 9). Finally, one student circled the entire phrase,
en de musubareru 'be strongly tied together', which was
predicted to be a difficult idiom (vocabulary feature 19).

This discussion indicates that the number of actual problems

is 7.
Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 3 1 33.3%
vecabulary 3 2 66.7%
Kanji 2 2 100%
Total 8 5 62.5%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 2 1 50%
vecabulary 3 2 66.7%
Kanji 2 2 100%
Total 7 5 71.4%

144

27. Ima nihon de wa, 'aruki' no gikggsha ga ggkilgg shi,
1 4

 

2
8
yaku kyuuhyakuman to mo iwarete iru.
4
12 9

'In Japan today, those who like to walk have increased
drastically and are said to number about 9 million.‘

One student marked the word, aikoo 'liking', and two
students marked the entire compound, aikoosha 'enthusiast',
indicating a problem with vocabulary. Five students marked
gekizoo 'drastic increase', which would mean that this
vocabulary item was difficult for the students. The fact
that the two Chinese characters, geki 'severe' and zoo
'increase', are difficult compounds the problem and adds two
more actual problems to the process of reading comprehension.
Finally, one student circled iwarete 'be said'. It seems
that the problem here is with the passive form of this verb
(grammatical feature 9). Thus, the number of actual problems
is 5. This is the first sentence of the ten 'change' as was
explained in the last chapter; however, this experimental
instrument does not appear to measure any interference at all
with reading comprehension. This could be just the
limitation of this instrument. Perhaps, rhetorical

interference can be measured at a much more global level.

145

Predictions

Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 3 1 33.3%
VOcabulary 2 2 100%
Kanji 4 2 50%
Ten 1 0 0%
Total 10 5 50%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 1 1 100%
Vocabulary 2 2 100%
Kanji 2 2 100%
Total 5 5 100%
1
28. "walk" to in senmon zasshi mo dete iru.

5
12 13
'There is now a special magazine, "walk".'

One subject circled the kanji, sen 'chief‘ in senmon
’specialty'. Five students circled zasshi 'magazine'. This
word is in the students' working vocabulary, but the Chinese
characters that compose this word are difficult: zatsu
'miscellany' and shi 'document'; therefore, this adds two
actual problems to the reading comprehension. It seems that

there are 3 actual problems in this sentence.

146

Predictions

Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 2 0 0%
vecabulary 0 0 n/a
Kanji 3 3 100%
Total 5 3 60%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar O 0 n/a
VOcabulary O 0 n/a
Kanji 3 3 100%
Total 3 3 100%
1
29. undoo buggkg o kgigngg suru tame.
1 4

2:
13
'To provide a solution to the lack of exercise.‘

One person circled undoo 'exercise', and two people
circled the whole phrase, undoo busoku 'lack of exercise',
indicating comprehension problems with both vocabulary items.
One person circled the kanji, kai 'understanding', indicating

a problem understanding it, and four circled the whole word

kaishoo 'solution'. Thus the actual number of problems is 4.

147

Predictions

Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 1 0 0%
vecabulary 3 3 100%
Kanji 3 1 33.3%
Total 7 4 57.1%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
VOcabulary 3 3 100%
Kanji 1 1 100%
Total 4 4 100%
2 2
30. fairyoku to ggiggigryoku no gggkgi ni igggg tame.
2 3 5 2
6 13

'To challenge one's physical and mental limits.’

Two students circled tairyoku 'physical strength'. Two
students circled the kanji, sei 'spirit', indicating a
problem with kanji, and three circled the whole word,
seishinryoku 'psychological strength'. Five students had
trouble with the word, genkai 'limit'. Two students had
trouble understanding the kanji in idomu 'challenge‘, while
two marked the whole vocabulary item. This means the total

number of actual problems is 6.

148

Grammar
vecabulary
Kanji

Total

Grammar
vocabulary
Kanji

Total

Predictions

Fulfilled
Predictions Predictions
2 0
4 4
5 2
11 6

Actual Problems

Actual
Actual Problems
Problems Predicted
0 0
4 4
2 2
6 6

31. figgigg no ikgggi no tame.
5 5

'To nurture the spirit.‘

Five students marked the word, konjoo 'spirit', and five
marked ikusei
marked, so it is difficult to tell what their problems with

kanji were.

probleun.

'nurturing'.

149

Prediction
Success
Rate

0%
100%
40%

54.5%

Actual
Success
Rate

n/a
100%
100%

100%

The individual kanji were not

As a result, there are only two obvious actual

Predictions

Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 1 0 0%
VOcabulary 2 2 100%
Kanji 4 0 0%
Total 7 2 28.6%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar O 0 n/a
Vocabulary 2 2 100%
Kanji 0 0 n/a
Total 2 2 100%
1 2
32. Arukitsuzukeru riyuu wa samazama da ga, aruki no ggiglgi
3 1 4
11 13 7

we yahari, aruku koto o tanoshimu koto daroo.
2 2 14

'There are various reasons for continuing to walk, but the
essence of walking is in its enjoyment.‘

One student marked the kanji, yuu 'reason, cause', in
riyuu 'reasons' (a kanji problem), and two circled the whole
word (a vocabulary problem). These problems were not
predicted. One student circled samazama 'various', also, a
vocabulary problem which was not predicted. Two circled zui

'path' (a kanji problem) in shinzui 'essence' while four

150

marked the whole word (a vocabulary problem). The result is

that there are 5 actual problems.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 6 O 0%
vecabulary 2 1 50%
Kanji 2 l 50%
Total 10 2 20%
Actual Problems
Actual Actual
Actual ‘ Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
VOcabulary 3 1 33.3%
Kanji 2 1 50%
Total 5 2 40%

33. Tanoshiminagara arukeba, kaze no iro ga miete kuru.
8 8 8

'If you walk while enjoying it, you can see the color of
the wind.’

In this passage no actual problems were indicated.
Therefore, no predictions were fulfilled. The fact that this
sentence is the ketsu or conclusion was explained in the last
chapter; however, this experimental instrument does not
appear to measure any interference at all with reading

comprehension. As mentioned before, this could be just the

151

limitation of this measuring instrument. As noted before,
perhaps rhetorical interference can be measured at a mmch

more global level.

Predictions
Prediction

Fulfilled Success

Predictions Predictions Rate
Grammar 3 0 0%
vecabulary 0 n/a
Kanji 0 0 n/a
Ketsu 1 0 0%
Total 4 0 0%

Actual Problems

Actual Actual
Actual Problems Success

Problems Predicted Rate
Grammar 0 0 n/a
VOcabulary 0 0 n/a
Kanji 0 0 n/a
Total 0 0 n/a

5.1.2. The Analysis of "Hoogen to Kyootsuugo". The
following is a sentence by sentence analysis:

1. Nihon wa, kuni ga semai war! n1, tochi ni yoru hoogen no
2 1

..L
13 13

chigai ga ookii.

‘For a small country, Japan has large differences in its
geographical dialects.‘

152

Two students circled wari 'proportion', and two circled
the phrase, wari ni 'in proportion to'. The fact that wari
was circled indicates that it is a vocabulary problem. The
fact that the entire phrase, wari ni, is circled indicates
that this phrase was not understood. Another student circled
hoogen 'dialect' to indicate vocabulary difficulty. Thus,

there are three actual problems.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 2 O 0%
Vocabulary 2 2 100%
Kanji 0 0 n/a
Total 4 2 50%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
VOcabulary 3 3 100%
Kanji 0 0 n/a
Total 3 3 100%
2
2. Nihon no chips; wa yama ya kawa ga ooku,.lotaloto
3

153

kootsuu ni fuben de aru ue, mukashi wa, nihyaku suujuu mo

1 LL
_1._

13 1
no daigzgg no ryoocbi ni wakarete sell; ga okonawarete
2 1 1 1

6 9 8 9

ita.

PAs for Japan's physical geography, in addition to there
being a lot of mountains and rivers naturally making it
difficult for transportation, in former times, Japan

was divided into 200 to 300 fiefs, owned and administered
by feudal lords.’

Two students marked the kanji, sei 'influence', in the
word izhisei 'typography' indicating a difficulty
understanding this kanji, and three marked the whole word
indicating a vocabulary problem. One student marked kootsuu
'transportation', indicating another vocabulary problem. One
circled the whole phrase, aru ue 'be in addition to', which
would indicate that there is, perhaps, a grammatical problem
here since aru functions as the verb of the relative clause,
motomoto kootsuu ni fuben de aru 'which naturally makes it
difficult for transportation', which modifies the relational
head noun, ue 'above'. This problem was predicted by
grammatical feature 13 (relative clauses). Another student
also circled run and e in this phrase, (a(ru ue)), indicating

a problem with word boundaries which may go back to a

vocabulary problem. Still another student just circled the

154

e. Clearly, this indicates a problem with word division and
vocabulary. Two students marked the word daimyoo 'feudal
lord', which would indicate a vocabulary problem. One
student also marked ryoochi 'territory', another vocabulary
problem. One student marked wakarete 'be divided', which
would perhaps indicate a problem with subordination (grammar
feature 8). Also, wakarete is in the passive which would
create a further grammatical problem (grammar feature 9).
Finally a student marked okonawarete 'be done', a passive

form (grammar feature 9).

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 7 3 42.9%
VOcabulary 6 4 66.7%
Kanji 4 1 25.0%
Total 17 8 47.1%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 4 4 100%
Vocabulary 5 4 80%
Kanji l 1 100%
TOtal 10 9 90%

155

3. Sore de, hitobito we, tagai ni jiyuu ni ikiki suru koto
4 2 2

7 2

ga muzukashikatta.

_1

4

'So it was difficult for people to freely associate with
each other.‘

Four students circled tagai 'mutual', two circled jiyuu
'free', and two circled ikiki 'association'. These are all
vocabulary problems. One student circled the segment of
text, to ga muzukashi, which is meaningless according to the
context. This indicates a problem with word boundaries which

may go back to a vocabulary or grammar problem.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 3 0 0%
VOcabulary 1 l 100%
Kanji 0 0 n/a
Total 4 1 25%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
vecabulary 3 1 33.3%
Kanji 0 0 n/a
Total 3 1 33.3%

156

4. Sono tame, kotoba mo, hitotsuhitotsu no chihoogoto ni

 

1
7
wekareru yoo ni natta no de aru.
l
9 13 2

‘For this reason, language, too, came to be divided into
individual geographical areas.‘

One student circled kotoba ‘language‘ indicating a
problem with the kanji in this word. Another student circled
the phrase, wekareru yoo hi ‘to be divided‘. It seems that
the difficulty in understanding this phrase is grammatical in
nature and has to do with the embedding (grammar feature 13 -
- relative clause) of the passive verb wekareru ‘be divided‘
(grammar feature 9). The passive form of wekareru seems to
be causing difficulty although this form was not explicitly

circled. Thus there are four actual problems in this

sentence.
Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate

Grammar 4 2 50%
Vocabulary 1 0 0%
Kanji 0 0 n/a
Total 5 2 40%

157

Actual Problems

Actual Actual
Actual Problems Success

Problems Predicted Rate
Grammar 2 2 100%
VOcabulary 0 0 n/a
Kanji 2 0 0%
Total 4 2 50%

5. Toku ni tooi chihoo to no ikiki we taihen na koto datta

2
1 6
kara, hyakunen hodo mae made we,.Aomgri no hito to
8 6
1
(gaggshila no hito ga deatte mo, hotondo kotoba ga
_JL__, 1 1
_1.,_

11 8
tsuujinakatta to in koto de aru.
4 12 2
‘Especially because association with a faraway district
was difficult, up to 100 years ago, even if a person from
Aomori and a person from Kagoshima met, they could not
communicate.

Two students marked ikiki ‘association‘. It seems this
is a vocabulary problem; however, it was not predicted.
Three students marked Aomori. More than a vocabulary
problem, this seems to be a cultural difficulty. In this

case, these students were not familiar with the geography of

Japan. Aomori is a prefecture in northern Japan. One

158

student marked the kanji, ka, in Kagoshima, and another
student circled both the Chinese characters, ka and go, in
the same word. Both of these characters are very difficult.
Three students circled the whole word indicating their
unfamiliarity with the word. Again, cultural understanding
is required. Kagoshima is a city and prefecture on the
southern island of Kyuushuu. One student circled the
gerundive verb form, deatte ‘meet‘. This seems to be a
grammatical problem because the gerundive verb form was
circled and the particle "mo" following the gerundive forms
the concessive construction: ...te mo ‘even if‘, which is a
form of subordination (grammar feature 8). Also deatte
‘met‘, is a verb compound (feature 11), which is analyzed as
follows:
de + atte
‘appear‘ ‘meet‘.

Finally, one student circled the vocabulary item, kotoba
‘language‘, indicating a problemnwith the kanji in this word.

This problem was not predicted.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 9 2 22.2%
VOcabulary 3 1 33.3%
Kanji 2 2 100%
Culture 2 2 100%
Total 16 7 48.3%

159

Actual Problems

Actual Actual

Actual Problems Success

Problems Predicted Rate
Grammar 2 2 100%
Vocabulary 3 1 33%
Kanji 4 2 50%
Culture 2 2 100%
Total 11 7 63.6%

6. Konnichi de mo, sono tochitochi no hoogen we, meinichi no

6

seikatsu no naka de ikiiki to tsukawerete iru.
2 1

1 9
‘Even today, the dialect of each area is being used
vividly in the middle of everyday life.‘

One student circled ikiiki ‘vivid‘. Another student
circled the passive tsukawerete ‘be used‘ (grammar feature
9). This is the ten ‘change‘ as was explained in the last
chapter (section 2.1). As noted before, this experimental
instrument does not appear to measure any interference at all
with reading <comprehension, which could be just the
limitation of this instrument. As indicated before, a better
solution might be to measure rhetorical interference at a

much more global level.

160

Predictions

Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 3 1 33.3%
vecabulary 1 0 0%
Kanji 0 0 n/a
Ten 1 0 0%
Total 5 1 20%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 1 1 100%
Vocabulary 1 0 0%
Kanji 0 0 n/a
Total 2 1 50%

7. FUdan yoku tsukau aisatsu no kotoba o mite.miyoo.
1 1

13 8
‘Let‘s look at words that are ordinarily well-used for
greetings.‘
One student marked fudan ‘ordinarily‘, a vocabulary
problem, and another marked kotoba ‘language‘, indicating a

problem with the kanji in this word.

161

Predictions

Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 2 0 0%
vecabulary 0 0 n/a
Kanji 0 0 n/a
Total 2 0 0%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
Vocabulary l 0 0%
Kanji 2 0 0%
Total 3 0 0%

8. "Konbenwe" to in yoru no aisatsu o, "Oban de gozaimasu"

1
12 13
"Oban ni narimashita" aruiwe, tada "Oban" to in tochi ga
1 1 1
6 12 13
aru.

‘There is a region where the evening greeting "Konban we"
(”Good evening") is said "Oban de gozaimasu", ”Oban ni
narimasu", or simply ”Oban".
One student circled Oban ‘evening‘ each time it was
used. Another circled aruiwa ‘or‘. These are vocabulary

problems.

162

Predictions

 

Prediction
Fulfilled Success

Predictions Predictions Rate
Grammar 5 0 0%
Vecabulary 0 0 n/a
Kanji 0 0 n/a
Total 5 0 0%

Actual Problems
Actual Actual

Actual Problems Success

Problems Predicted Rate
Grammar 0 0 n/a
VOcabulary 2 0 0%
Kanji 0 0 n/a
Total 2 0 0%

1
9..Mata, "Arigotoo" no kaweri ni, "Ookini" to ka "Dendan" o
_.1_
1
7 6

tsukau chihoo ga aru.

13

"Again, there is a district which uses "Ookini" and
"Dandan" instead of "Arigatoo" ("Thank you").

One student circled the kanji, ka ‘exchange‘, indicating
difficulty in understanding it. One student circled the
whole word, kawari 'substitute‘, indicating a vocabulary
difficulty. Another student circled weri ni, which has no

meaning in this context. Clearly, this is a boundary problem

163

which has its roots in vocabulary difficulty. As a result,

there are three actual problems in this sentence.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 3 0 0%
VOcabulary 0 0 n/a
Kanji 0 0 n/a
Total 3 0 0%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 0 0 n/a
Vocabulary 2 0 0%
Kanji 1 0 0%
Total 3 0 0%

10. KOrera no aisatsu kotoba ni we, sono tochi sono tochi no
1

1
aji ga kamotte iru.
2

‘As for these words of greeting, they are full of the
flavor of each particular region.‘

One student circled kotoba ‘language‘ indicating a
problem with the kanji in this word. The same student

circled the Chinese character, aji 'favor‘. It seems that

164

this is a problem in understanding the meaning of this kanji.

Two students circled the gerundive form, komotte ‘be filled

with‘. This would be a very difficult word for them, so this

seems to be a vocabulary problem.

Predictions
Prediction

Fulfilled Success

Predictions Predictions Rate
Grammar 3 0 0%
Vocabulary 1 1 100%
Kanji 0 0 n/a
Total 4 1 25%

Actual Problems

Actual Actual
Actual Problems Success

Problems Predicted Rate
Grammar 0 0 n/a
Vocabulary 1 1 100%
Kanji 3 0 0%
Total 4 1 25%

11. Nazenara, hoogen wa, sono

 

tochi no seikatsu ya raglan;

.3..-

_1__
1:
7
2
to fukai tsunagari ga ari, kazokudooshi,
3
l 8

165

to-odacbidooshi, tochi no hitodooshi ga kutsuroida

 

1 _§;_
6 13
2
hanashi o suru toki ni we,shizen ni [kuchi o tsuite] dete
1
l3 8 8

kite, tagai ni yoku kimochi o tsuujiau koto ga dekiru

8 2 13

mono dakara de aru.
8

‘The reason for this is a dialect has a close relation

with the life and history of the local area, so when

family, friends, and people of that region engage in
relaxed conversation, the dialect just rushes out
naturally and is able to mutually convey emotion.‘

One student underlined the phrase nazenara, hoogen we ‘the
reason is a dialect‘. The sentence level conjunction,
nazenara ‘the reason is‘, connects this sentence with the
previous explaining why a dialect is so closely connected
with the region it is associated with. It seems that the
problem here is with the sentence connector, nazenara
(grammar feature 7). This person is also apparently having
trouble with the vocabulary item, hoogen ‘dialect‘, along
with another student who circled this item. Two students
circled the Chinese character in fukai 'deep‘. Three

students circled tsunagari ‘connection‘ indicating’ a

vocabulary problem since this word is written in hiragana in

166

this text. Students circled the dooshi ‘fellow‘ in
kazokudooshi ‘fellow family members‘, tomodachidooshi ‘fellow
friends‘, and hitodooshi ‘fellow citizens‘, which clearly
indicates a vocabulary problem. One student circled the
whole word hitodooshi, again revealing a comprehension
problem with the dooshi compound. Six students circled the
verb kutsuroida ‘to be relaxed‘. It seems that this is a
problem with vocabulary. Five of the students circled
kutsuroi, and another student circled kutsuro, which
indicates they did not know the word well enough to know the
proper word boundaries. It seems what the students saw here
was an adjective or a noun but not a verb. Actually, the
fact that kutsuroida is a plain past tense verb that modifies
the head noun, hanashi ‘speech‘ could have confused the
students (grammar feature 13 -- relative clauses). To
continue with the analysis of this sentence, two students
circled the kanji, zen ‘nature‘ (a kanji problem), in shizen
‘natural‘ while one student marked the whole word (a
vocabulary problem). Three students marked tagai ‘mutual‘,
which indicates a vocabulary problem. This word was also
marked in sentence 3 of this passage. The fact that this
sentence is the ketsu or conclusion was explained in the last
chapter. As mentioned before, this experimental instrument
does not appear to measure any interference at all with

reading comprehension on the part of the ketsu. As noted

167

before, this could be just the limitation of this measuring
instrument. Maybe rhetorical interference can be measured at

a much more global level.

Predictions
Prediction
Fulfilled Success
Predictions Predictions Rate
Grammar 13 2 15.4%
VOcabulary 7 5 71.4%
Kanji 2 0 0%
Ketsu l 0 0%
Total 23 7 30.4%
Actual Problems
Actual Actual
Actual Problems Success
Problems Predicted Rate
Grammar 2 2 100%
vecabulary 8 5 62.5%
Kanji 2 O 0%
Total 12 7 58.3%

5.2. Conclusions. This section mainly presents a summary
and some preliminary discussion of the data. Further
expanded discussion of the implications of this data is
presented in section 7.2.1. 1

5.2.1. Conclusions for "Aruke Orinpikku". Table 5 contains
a summary of the results (refer to chart 3 for a graphical
picture) for the predictions in "Aruke Orinpikku“. It

contains the predictions and the actually fulfilled

168

predictions for each sentence. The fulfilled predictions are
the intersections between predictions made in the last
chapter and the actual problems the students encountered in
this experiment. According to table 5, only 24.71% of the
grammar predictions were fulfilled. The subjects in this
experiment mainly circled words and not larger units, so it
is hard to determine if they knew how the words fit together.
Since grammar is the study of how words fit together, it is
hard for this empirical study to determine the effect of
grammar on difficulty. The low prediction success rate for
grammar indicates this. The rate of fulfilled predictions
for vocabulary and kanji is much higher: 77.08% and 62.90%
respectively. It seems that problems with vocabulary and
kanji are much easier to predict, perhaps, because this
instrument can measure these word level entities more
readily. No predictions for rhetoric were fulfilled. This
instrument had trouble measuring rhetorical problems as
explained in relation to sentences 20 and 33 of “Aruke
Orinpikku”. Also, the one cultural prediction in the passage
for sentence 13 was not fulfilled. The total prediction
success rate for this passage was 48.99%, due mainly to the
low prediction success rate for grammar. It seems like this
empirical investigation had trouble measuring grammatical
problems since the subjects just circled words and not larger

segments of text. Computer software can be designed to

169

Table 5

Aruke Orinpikku Predictions

ulfill. ulfill. ulfi l. ulfill. ulfill.
amm rammar ocab. ocab. i an i hetori hetori ltur ltur
t. redict. redict. redict. redict. redict redict. redict. redict. redict redict

1 1 1 1
0 0 0
1 1 1 1
1 1
1 1

0AA

ro--—-.ur\iooiNu—ooi
ouomodoofio
---oo—-oo-to--

0
0
1

8

rd. 8 = 4.71
otal Ful illed Predictions a
Total Predictions =

Total Prediction Success rate =

 

170

Number 01 Predictions

Aruke Orinpikku Predictions Chart

90

80

70

60

50

40

30

20

10

 

Grammar

 

 

Vocabulary

 

Kanji

Chart 3

 

U a
L

o a
u .—
0.) 3
-C U
0:

171

 

 

I Predictions

1:1 Fulfilled Predictions

 

 

measure more fully the contribution of grammar to passage
difficulty by dividing the text into larger segments. See
chapter 6 for a fuller discussion of the computer program.

Table 6 contains a summary of the actual problems (refer
to chart 4 for a graphical picture of this summary) the
subjects had with "Aruke Orinpikku". Here the grammatical
prediction rate for the problems the students actually had
was quite high--90.9l%. This means the grammatical problems
the students actually had with grammar were predicted
successfully. This is to be expected because so many
grammatical predictions were made. However, it seems like
this instrument missed a lot of these predictions (see the
previous paragraph). The computer program attempts to
further measure the effect of grammar on textual difficulty
(see chapter 6).

Given the actual grammatical problems as determined by
this experiment, it is possible to rank the grammatical
features to determine the order of their actual influence on
difficulty. The ranking for the grammatical features in
terms of the magnitude of their actual influence is
(excluding features 5, 17, 18, 19, and 20):

EQQEQIQ. Humber_2f_£rehlems.

(1) Passive (9)

(2) Tense (4)

(3) Subordination (8)
(4) Causative (10)

(5) verb complexes (11)
(6) Quotations (12)

NNNIFbIhQ

172

Table 6

Aruke Orlnpikku Actual Problems

Actual Actual Actual
ctua Grammar ctual Vocabula Actual Ken 2'
Grammar roblems Vocabu roblems Ken 1' Problems
ntence Problems redicted Problems redicted Problems Predicted

o‘dd

O-OO-

O‘ONO-‘OO-s

O
NNOA-s-‘NANONNOWO‘

1
0
1
1
1
2
1
1
2
1
0
0
1
0
0
1
1

oo—‘NO—‘o—‘OOOONO
.5

ddNNN-‘O

N
N

d

018 S

ual = 90.918
ota Actual Prob ems Predicted =
otal Actual rob lems =

otal Actual Success Rate =

 

173

Number of Actual Problems

Chart 4

Aruke Orinpikku Actual Problems Chart

60 —~

50

4o 1

 

I Actual Problems

30 4r DActual Problems

Predicted

 

 

 

20 ~~

 

 

 

 

 

Grammar
Vocabulary
Kan ji

174

(7) Relative clauses (13)

(8) Homophonous particles (1)
(9) NOun head modification (6)
(10) Nominalizers (2)

(ll) Anaphora (3)

(12) Conjunctions (7)

(13) Hedges (14)

(14) Further embeddings (15)
(15) Comparatives (16)

OOOOOOOOH

According to the results of this experiment, the passive (9)
has the most influence on difficulty since it represents 7
problems. Next comes tense (4) with 4 problems,
subordination (8) with 4 problems, and so on. What this says
is that for this passage, the passive voice is a more
important determiner of difficulty than tense and
subordination. These are observations that need to be
further tested empirically during the experiment involving
the computer.

To continue with the summary of the actual problems
(contained in table 6), the vocabulary and kanji prediction
rates for the actual problems were rather good at 62.71% and
67.24% respectively. The total actual success rate was
69.06%, which means that there was a fairly good prediction
success rate for the problems the subjects actually had.

At the end of last chapter, certain predictions were
made about the relative contributions of grammar, vocabulary,
kanji, rhetoric, and culture to the total number of problems.
The predictions were: 1) grammar -- 42.71%; 2) vocabulary --
24.62%; 3) kanji -- 31.16%; 4) rhetoric -- 1.01%; and 5)

culture -- 0.50%. According to the results of this

175

Chart 5

Percentage of Actual Problems in 'Aruke
Orinpikku'

Grammar
15.8358

  
 
 

Kanji 41.7358

Vocabulary
42.45%

176

experiment, each area made the following contributions to the
total number of actual problems: 1) grammar -- 15.83%; 2)
vocabulary -- 42.45%; and 3) kanji -- 41.73%. A chart of
these findings for the actual problems is shown in chart 5.
It seems like the importance of grammar was over-predicted
and the importance of vocabulary and kanji under-predicted.
However, this could be an artifact of the inadequacy of this
testing instrument, especially in the area of measuring
grammatical difficulty as discussed above. Furthermore,
problems with rhetoric and culture were not measured in the
results of this experiment. The computer software explained
in the next chapter attempted to overcome these shortcomings.
As a measure of the actual difficulty of this passage,
the total number of actual problems (139) can be divided by
the number of words in the passage (329), which comes out to
.42. Also a total of 294 circles or marks were made in this
passage during the experiment. If this is divided by the
number of words in the passage, the result is .89. This can
also be another measure of difficulty.
5.2.2. Conclusions for "Hoogen to Kyootsuugo“. Table 7
summarizes the results (refer to chart 6 for a graphical
picture) for the predictions in ”Hoogen to Kyootsuugo". The
predictions and the actually fulfilled predictions for each
sentence are contained in this table. The fulfilled

predictions are the intersections between predictions made in

177

Table 7

to Predict

Fulfill. Fulfill. Fulfill. Fulfill. Fulfill.
Gramm Gramm Vocab. Vocab. Kani Kani Rhetori Rhetori Culture Cultur
t. Predict. Predict. Predict redict. redict. Predict. Predict. Predict. Predict Predict

9

‘OOOO—OOOOO

NOOOO-‘NNO
(fl-OOOOO-‘O-‘A
0000000000

3
4
9
3
2
5
3
3
3

11

1

Total 54

0

red. Succ. 2 = l .
ota u fil Predictions

Total Predictions

Total Prediction Success rate =

 

178

Number of Predictions

Chart 6

Hoogen to Kyootsuugo Predictions Chart

60 1—

 

. Predictions

1:1 Fulfilled Predictions

 

 

 

 

 

Kanji
Rhetoric
Culture

L

co 6

E m
D

g o

L 8

o o
>

179

the last chapter and the actual problems the students
encountered in this experiment. According to this table,
only 18.52% of the grammar predictions were fulfilled. As in
the previous passage, the subjects in this experiment mainly
circled words and not larger units, so it is hard to
determine if they knew how the words fit together or if they
knew how to indicate syntactic difficulties from the
discourse level on down. As mentioned previously, since
grammar is the study of how words fit together, it is hard
for this empirical study to determine the effect of grammar
on difficulty. The rate of fulfilled predictions for
vocabulary is much higher at 63.64%. Again vocabulary
problems seem easier to predict as in the case of the
previous passage. There is a 37.5% success rate for the
kanji predictions. Eight problem areas for Chinese
characters had been predicted, but only three were circled.
Again as in the previous passage, no predictions for rhetoric
were fulfilled. This seems to be a limitation of this
experimental instrument as explained when sentences 6 and 11
of this passage were discussed. However, in this passage,
the two cultural predictions were successfully fulfilled. It
was easy to predict that the geographical names, Aomori and
Kagoshima, would cause difficulty since English-speaking
students would probably not be familiar with these place

names. The total prediction success rate for this passage

180

was 32.95% due to the low prediction success rates for
grammar and kanji. As with "Aruke Orinpikku", it seems like
this empirical investigation had trouble measuring
grammatical problems since the subjects mainly circled words
and not larger segments of text. As mentioned previously in
the discussion of the results of "Aruke Orinpikku", computer
software can be designed to measure more fully the
contribution of grammar to passage difficulty by dividing the
text into larger segments. See chapter 6 about the computer
program.

Table 8 contains a summary of the actual problems (refer
to chart 7 for a graphical picture) the subjects had. The
grammatical prediction rate for the problems the students
actually had was 100%. A lot of grammatical predictions had
been made in the previous chapter, so it is not surprising
that the actual grammatical problems the subjects had were
well covered.

Given the actual grammatical problems as determined by
this experiment, it is possible to rank the grammatical
features in "Hoogen to Kyootsuugo" to determine the order of
their actual influence on difficulty. The ranking for the
grammatical features in terms of the magnitude of their
actual influence is (excluding features 5, 17, 18, 19, and

20) (see page 184):

181

ctua
tual ammar tua

ammar robl. ocab.

ntence Prob . redict. robl.

OO-‘NNOAO

018 8

us Success

1 rob redicted =

otal Actual Problems

Total Actual Success rate =

0
3
l
l
2
2
l
8

Table 8

Actual Problems

tua ctua
ocab. tua an i tual
rob . an i robl. lture
redict. robl. redict. robl.

182

ctual
ulture
robl.
redict.

 

Chart 7

Hoogen to Kyootsuugo Actual Problems Chart

Number of Actual Problems

30

2'5

20

15

10

 

Grammar

 

 

Vocabulary

 

 

 

Kanji

183

 

 

I Actual Problems

1:1 Actual Problems
P redicted

 

 

 

 

 

Eéfliflifl.

(1) Passive (9)

(2) Relative clauses (13)

(3) Subordination (8)

(4) Conjunctions (7)

(5) Verb complexes (11)

(6) Homophonous particles (1)
(7) Nominalizers (2)

(8) Anaphora (3)

(9) Tense (4)

(10) NOun head modification (6)
(11) Causative (10)

(12) Quotations (12)

(13) Hedges (14)

(14) Further embeddings (15)
(15) Comparatives (16)

OOOOOOOOOOHHNWh

According to the results of this experiment, the passive (9)
has the most influence on difficulty since it represents 4
problems. Next comes relative clauses (13) with 3 problems,
subordination (8) with 2 problems, conjunctions (7) with 1
problem, and verb complexes (11) with 1 problem. What this
says is that for this passage, the passive voice is a more
important determiner of difficulty than conjunctions and verb
complexes. These are observations that need to be further
tested empirically during the experiment involving the
computer.

To continue with the summary of the actual problems
(contained in table 8), the vocabulary prediction rate for
the actual vocabulary problems was 51.72%. This rate
indicates that they had a fair amount of trouble with
vocabulary items that had been predicted not to cause
trouble. The kanji prediction rate for actual kanji problems

was 20%. They seem to have had a large number of problems

184

with kanji in areas other than predicted. The prediction
rate for the actual cultural problems in this passage was
100%. This is due to the successful prediction of problems
related to understanding the place names, Aomori and
Kagoshima. The total actual success rate for this passage
was 54.39% especially due to the low prediction success rate
for kanji.

At the end of the previous chapter, certain predictions
were made about the relative contributions of grammar,
vocabulary, kanji, rhetoric, and culture to the total number
of problems in this passage, "Hoogen to Kyootsuugo". The
predictions were: 1) grammar -— 61.36%; 2) vocabulary --
25%; 3) kanji -- 9.09%; 4) rhetoric -- 2.27%; and 5) culture
-- 2.27%. According to the results of the experiment on
”Hoogen to Kyootsuugo“, each area made the following
contributions to the total number of actual problems: 1)
grammar -- 19.30%; 2) vocabulary -- 50.88%; 3) kanji --
26.32%; and 4) culture -- 3.51%. Chart 8 illustrates these
findings. It seems like the importance of grammar was over—
predicted. However, again this could be an artifact of the
inadequacy of the testing instrument used for this passage,
in particular in the area of measuring grammatical difficulty
as discussed above. The contribution of kanji to the total
number of problems was predicted to be 9.09%, but according

to the experiment it was 26.32%, which indicates that the

185

Chart 8

Percentage of Actual Problems in "Hoogen to
Kyootsuugo'

Culture 3.512;

Grammar
19.30%

 
 
  

Kanji 26.32%

Vocabulary
50882

186

importance of kanji was under-predicted. Problems with
rhetoric did not appear in the results of this experiment;
however, cultural problems did, constituting 3.77% of the
total number of problems, which is very close to the
predicted value, 2.22%.

As a measure of the actual difficulty of ”Hoogen to
Kyootsuugo", the total number of actual problems (57) can be
divided by the number of words in the passage (234). This
comes out to .24, which is lower than the corresponding
figure for "Aruke Cninpikku", which was .42. This means
"Aruke Orinpikku" is definitely the more difficult passage.
Also the subjects made a total of 86 circles or marks in
"HOogen to Kyootsuugo" during the experiment. If this number
is divided by the number of words in this passage, the result
is .37, which is lower than the corresponding figure for
"Aruke Orinpikku", which was .89. This further indicates‘
that "Aruke Orinpikku" is the more difficult passage. This
gives empirical validity to the prediction that "Aruke
Orinpikku" is the more difficult passage.

The question that remains is whether the subjects in
this experiment really understood these passages. Because of
this, the researcher has developed software to help determine
the kind of help students need in order to comprehend a
Japanese text. This software attempts to motivate students

to comprehend the text by providing comprehension questions

187

that must be answered by reading the text carefully. This
software also attempts to determine the effect of grammar on
readability by keeping track of how much the students use the
grammar help facility in the program. In this way, perhaps
some of the shortcomings of the experimental technique
covered in this chapter can be compensated for. At this
point, it is difficult to determine to what extent the
students comprehended the passages discussed in this chapter.
As a result, based on the data and analysis in this chapter,
software has been developed to help determine the level of
the students‘ reading ability and to help them understand

these texts.

188

CHAPTER 6

The Computer Experiment

6.0. Introduction. A HyperCard computer program was
developed for the two passages, "Aruke Orinpikku" ("Walking
Olympics”) and ”Hoogen to Kyootsuugo“ ("Dialects and the
Common Language"). Assistance was provided by having the
reader choose the information that is necessary to understand
these passages. This information was divided into five
different areas: vocabulary, kanji (Chinese characters),
grammar, rhetoric, and culture (background knowledge). The
content of this information was determined both by the
hypotheses as to what contributes to textual difficulty as
discussed in chapter 4 and by the empirical study of what
textual elements the students themselves thought were
difficult as presented in chapter 5. Beyond this, this
program could also record the information the reader used to
understand these texts. It could keep track of how often
readers made selections on vocabulary, Chinese characters,
grammar, rhetoric, and culture in order to understand these
passages. This information was analyzed to determine what
elements in the passage made the text difficult. The
assumption was made that the more difficult textual elements
needed to be accessed more in order to understand the text.

The data that was contained in the readers‘ records of how

189

often they used information to understand the passages was
analyzed to acquire an overall picture of the sources of
textual difficulty. When compared to the study in chapter 5,
this was a different kind of empirical study of what the
readers found difficult in these passages. In fact, this
program was based upon the results of the study contained in
chapter 5 which acted as a preliminary pilot study around
which this program was developed. In summary, this program
not only let the readers choose the help they needed to
understand the passages but also analyzed the difficulty of
the passages based on the help they received while processing
the texts. This produced an empirical analysis of the
sources of textual difficulty.

6.1. The Program. An Apple Macintosh IIsi was used to
develop the program and test it on-line with students. The
Macintosh model used in this experiment has a standard 80MB
hard disk drive with 5MB of internal RAM. The video display
was on a l3-inch high-resolution monitor positioned at eye
level for the readers who were sitting in front of the
monitors. A mouse was also utilized to enable the reader to
interact with the computer screen. As mentioned before, the
programming environment for this program is HyperCard, an
authoring environment developed by Apple Computer. The
HyperCard implementation used in this study runs on the Apple

KanjiTalk System, version 6.0.7, a system which utilizes the

190

 

Japanese writing system. As mentioned in chapter 2, section
2.4, HyperCard employs a hypertext environment which allows
the user to choose words or linguistic units of larger size
on the screen in order to have the program define, translate,
or explain them. These words are chosen with the aid of a
mouse which is part of the computer environment. The reader
does not interact with the computer by using the keyboard.
All the interaction is done through the use of the mouse,
which makes it more convenient for the user. In summary, the
purpose of this environment is to facilitate the
comprehension of Japanese in a way that is natural and
intuitive to the reader. The following are the sections of
the program that the subjects in this experiment could
utilize.

6.1.1. The Introduction to the Program for the User. When
the subjects go through the introductory section to the
program, they learn how each page of the Japanese text is
organized. An illustration of how a page of Japanese text,
in this case "Hoogen to Kyootsuugo”, is presented is shown in
Figure 1. The subjects receive the explanation that the
screens with text have buttons on them which can be clicked
with the mouse to provide the help that is needed to
understand the text. They are told that there are six kinds
of buttons: vocabulary, kanji, grammar, rhetoric, culture,

and text. The subjects learn that when the yggapglg;y_button

191

Figure 1

 

75% t: #333!

:5 amu.umetubuu.increasesquvxeu.are

:; 1119111 melting. ttthEigCTfl‘t‘bbax. new. :Efl
:-+sa>xsa>o:mceentasvfibntun. tint. Aau. it:
:2;micamcfierteztmcrawak. force) 815 --o—
;: 503119.75 tram-tantra): 011:6. t<tziauitll75t0i58

glufctmhfxztfc‘ arms stature. fifimAtEfilo
guide-‘4 are newsreaueear; more? 356

.; ears toifitfiwfitu IBOiiSCDIP‘t‘ESIStiIbh
grub. ewe/cumastxeeosxtnrbxa. [Chit/cm tin
3.5fiaamiaot. rsukrramtn ifstthlznwtbfclbbu
:41. tar: Halt/c1 t§5iflflb§. tic. room—'21 within;

;: [15381:] Ur fraud/u tflafififlbb. meanings:
5‘11. {mistoieomezearua._rests. 75311. «to-1:190)
gfifii’t’bflfitmiomfllflbv. 13851... was». :UMJAE
4507f<obumitfbt 81.11 am: Dtoti‘t 18131. fan‘u
2;:1:.1:(MSEHUS’JEtfl‘CSbbok‘vBI‘bb.

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO
......................

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
......................

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

 

I IIll-l-ItrIl‘lflIrilillilJ-I.I‘I-lllI-j‘ll-II-IIII-IJII'----

.....................
OOOOOOOOOOOOOOOOOOOOOO

 

PRESS THE VOCABULARY,KANJ|,

IN THESE RESPECTIVE AREAS

 

AND GRAMMAR BUTTONS TO GET HELP ‘

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
......................
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

......................
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
......................

 

 

O
..............................................................
.............................................................

192

......................
OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOOO

is clicked with the mouse, underlines appear in the text
which mark the words that are predicted to be difficult for
them as determined by the analysis of chapter 4 and the
empirical study in chapter 5. These words can be clicked on
to get the necessary help. They also learn that by clicking
the kanji, grammag, gngtgrig, and gulgg;g_buttons, they can
receive help in those respective areas. When they click the
(kanji_button, difficult (as determined by chapters 4 and 5)
kanji are underlined, and those can be clicked on to get the
appropriate help. When the grammar_button is clicked, parts
of sentences are underlined to indicate segments which are
predicted to be difficult for this group of subjects (based
on the studies of chapters 4 and 5). When the rhgtgrig,
button is clicked, sentences that mark the ki 'beginning‘,
shoo ‘development‘, ten ‘change‘, and ketsu ‘conclusion‘ are
underlined. When these sentences are clicked on,
explanations about this rhetorical structure are given to the
subjects in this study. When the ggltgrg button is chosen,
sentences or sections that present cultural difficulties are
underlined. When these sentences are selected, cultural
explanations are given based upon the predictions of cultural
difficulty in chapter 4 under feature 20. Finally, the text,
button just gives the text without any underlining to
distract the reader. In summary, these six buttons give the

reader a variety of options by which the text can be engaged

193

Figure 2

 

81:13.!b'12'2m s fitzunosgyuqmts. ate
a“ memm kmcxerasm new: rate

coexteceenuamak one. a < mum-em

a s. at senescence water“.

ear . winter) 1:. saofioerfiaabn
i:

. fck' 115.1161.) tiiiifllfl‘tb. 2k. L‘Jfi‘t‘ii 011119.15...
;. 81381:] t1? ”fluff/v.1 {1151112572136 ChBObL‘éOELC
£11 toimtoimomemwa. K‘E’IB. 11. 4105:1110)

emeesme—I . w—

Oéfiflkfifihffliflhbfl‘fl‘ft {11.1. A011 fL:
6Ztil‘6‘1‘irbtl‘of: {with 218—9—-

ufctgmkfiztfé was steamers momma 1'

Swamuaasu continua. z. wow 3:
:;5Eomiaot. rm rerun: r~ titanium;

3: ifi‘t’flitflumvb 0 155m this HOAK.

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
......................
OOOOOOOOOOOOOOOOOOOOOO

fifififififififififififififififififififififi

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
......................
OOOOOOOOOOOOOOOOOOOOOO
..................

OOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

 

‘If-IliirI-ljl‘r-II-III --lll-IIJ- ‘lll‘it11--------‘Ijl

OOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

 

 

PRESS THE UNDERLINED WORDS
TO GET VOCABULARY INFORMATION

 

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
QQQQQQQQQQQQQQQQQQQQQQ
......................
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO
QQQQQQQQQQQQQQQQQQQQQQ
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

 

 

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

194

in order to facilitate textual understanding.

6.1.2. The Vocabulary Section. As mentioned above, when the
subjects click the W button with the mouse,
underlines appear in the text under the words and phrases
that are predicted to be difficult for the subjects. These
words have been determined to be difficult by the analysis of
chapter 4 and the empirical study in chapter 5. Figure 2
contains the picture of the computer screen that appears for
”HOogen to Kyootsuugo" when the vocabulary button is clicked
by the mouse. Notice that problematic phrases have been
underlined. For example, in sentence 1, the phrase, weri ni
‘in proportion to", and the word, hoogen ‘dialect‘, have been
underlined. By clicking on the phrase, weri ni, with the
mouse, a window appears at the bottom of the screen that
defines the meaning of this phrase (see Figure 3). This
window contains this phrase written both in hiregena and
Roman letters along with the definition and the concatenation
of word categories (noun + particle) contained in the phrase.
Another example is the definition that appears in the window
at the bottom of the screen when hoogen is clicked on (see
Figure 4). Here the window contains this word written in
Chinese characters, hiragana, and Roman letters along with
the meaning and the word category, noun. The purpose of this
information on vocabulary is to help the reader comprehend

text in a way that is both intuitive and easy to understand.

195

Figure 3

 

Etfizfim‘ 5’

233.2411 lflttumu; but: tbfimszfiufifxsu. Eta)
2:39.511 lli’PllIfJ‘§< btbtflk 1:311? bola. 672191123!!!
:;+ti0) 0.91.5111?- fitf‘ 111116371"?thme ‘th‘t‘ gnu“: 7*;_‘

.; are I
3 knffiuhl 1&25 iflfibb. 2n
satultfiikhkhltflfiflfiflbo
gu w Mist ezgo Iub. use?
gees eneeauofl vesu

gfi11§§g¥§ffatemrea§&1#§¥%%g

   
  
   

 

ibuu

f; wari ni

Meaning: For, despite

  
   
  
 

 

 

 

<Noun + Particle>

196

 

Figure 4

 

Etm

131121: sztb‘ta‘ffl'bii‘ofc. 10ft!) at».
31903111753: menu: 13:51.: 0156. t < intuitefitfliiz.

25E®bu of
:3 . mart-2111.1 £25 itwibfb. tic. 601113.! 03%.
;. 83581:] Ur Hilde/cl tfl5fififlbe. :htobuaoi 1:
$11. toimtoimommmua. rune. £111. «1033:1190
-: maemtmumew.

2111éeeeseftf 111.

81511 Ifl‘fi'tl.‘ ". 11111:.taziaosmwxsu.ato
11 lll’PIlIfi‘;< £1th Tilfbbij. 0111.1: :53
01_Kfiflhffilifififibhtb\k. {hr A411 TL:

—3— -

117c1‘flhttltfcofcfl‘5.EfifiElfit‘t‘u mas/tamed) 23

if b. lit/1.8 #3611115): tl.\5:t‘(‘1:b. .

.: #B‘t‘ . «10111912111109 ‘ECDIIEOFPTWJIbh 3:

I-‘Tué. n " 1(if5bl. \ 00) wagjjl‘tbl’). 21A. 11.11] t1) :
‘ I?) Cflfltbkl

b.5112:
‘ -

11312511 Ifc5£2l¢ :IdIWJAE.

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO

..........
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

..........
OOOOOOOOOOOOOOOOOOOOOO
......................

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
QQQQQQQQQQQQQQQQQQQQQQ

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOO

 

lili-lr-tI-‘ii-Ill-l-IllIl-lr-.I‘rn-lI-‘I-I-l-l

 

 

$21111111r*:3111..

35 a
:3 11 5 11A.
:3 hoogen

3} Meaning Dialect
<Noun)

 

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
......................
OOOOOOOOOOOOOOOOOOOOOO
......................

 

OOOOOOOOOOOOOOOOOOOOOO
QQQQQQQQQQQQQQQQQQQQQQ
......................
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

 

 

A
O iii-‘ICI-IICIIr‘Illi'lll‘lr‘lj‘lfiliilliir‘n IIIIIIIIIIIIIIIIII

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

197

 

OOOOOOOOOOOOOOOOOOOOOOO

In summary, the focus of the kind of help that is given here
is on the problems that this group of subjects were having in
understanding the text, based on the studies of both chapters
4 and 5. This program builds upon the knowledge the subjects
already have of Japanese vocabulary and extends that
knowledge when requested by providing vocabulary help. The
subjects have the power to choose what vocabulary help they
need and when they need it. This potentially creates a user—
friendly environment.

6.1.3. The Kanji Section. As with the vocabulary, when the
subjects click the Kanji button, underlines appear in the
text under the kanji that are predicted to be difficult for
the subjects as determined by chapters 4 and 5. When the
subjects click on the underlined Chinese characters, they get
appropriate help to understand the kanji that are difficult
for them. When a Chinese character with an underline is
chosen, the window below the text shows information about
this character. The on (Chinese pronunciation of the
character) is given in capital letters, while the kun
(Japanese pronunciation of the character) is given in small
letters. The meaning of the character is also provided in
the window. Figure 5 contains an example of how this
information is presented in the window below the text. The
noun, moyooshi, appears in the first sentence of "Aruke

Orinpikku". The noun itself means 'event'. The Chinese

198

Figure 5

{E SAI

moyoo(su) to hold (a meeting)

199

character in this word is given in Figure 5. This character
has a Chinese pronunciation of $111, which is used in
compounds, and a Japanese pronunciation of moyoo(su), a verb
which literally means 'to hold (a meeting)’. The stem form
of moyoo(su) is moyoo(shi), which is the noun which appears
in this text. In summary, the purpose for providing
information on Chinese characters is to provide in-depth
understanding of word meaning. The use of Chinese characters
forms an important component of Japanese vocabulary that the
reader of Japanese needs to be aware of.

6.1.4. The Grammar Section. As seen in Figure 6, when the
g;amma;_button is clicked, parts of sentences are underlined
to indicate segments that are determined to be difficult for
this group of subjects (based on the studies of chapters 4
and 5). When these segments are chosen by the subjects,
grammatical information about these segments appears in the
window at the bottom of the screen. Figure 6 contains the
text for page 1 of "Aruke Orinpikku". The clause, ooki na
moyooshi ga okonawareru 'a big event is held', has been
selected, and information about the passive verb, okonawareru
'is held', appears in the window at the bottom of the screen.
The passive verb in this clause has been explained in this
window because the passive construction was predicted to
cause difficulty in section 4.1.9 of chapter 4. The fact

that a student marked okonawareru in the empirical experiment

200

Figure 6

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

 

I’vblhpn

 

p.»-pr...-npu.pnnnn»p-npnv-v
......»1nnI-nnn-11 -

 

(WM amp;

1511 My

12229¢§Efl§.3§43bflfi§pfl§. 71—f1-

U114) I“ ”I Egg“ W”

Lililflifc'
fi°1.11871<71’bi+nkfl‘¥mbk. $111] 71903 ¥E- 3
W

fiEmtfiflflfifiIUOL. rtfiufiztwig
M

1‘ J.
Lid—3km fi ?3 A11 We don‘t want toflg t1‘2‘ fififimta
45W-

Ev?

|I|DI|I
. ...
»
_¢
'0
x
..
-
u.

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

Knnll

IQIlIIIDI
u...
, ,:
Ifihlép

10'15'4 X —’\;?1:;

   

A.
1W1] t
m

B

IIIIDIIDIIIDI

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

 

pn-pn-pnnn-nn»:n.r-¢vv-)

'l'hvlri, fi'

 

0.5).».-DIIODIDIIIIIODIIDIDII

’Ibhb. is I]. p-ficolfir‘5.

 

 

 

 

 

 

 

 

201

of difficulty (as explained in chapter 5) indicates the
possibility that the passive form of this verb is a source of
difficulty. Therefore, the fact that this verb is in the
passive is explained to the users of this program. If a user
of this program utilizes this information in trying to
understand the text, that fact is recorded by the computer.
Also, in this case the computer records the fact that
information about grammatical feature 9 (passives) was given
to the user of this program. In summary, the grammatical
help section of this program takes into consideration both
the hypotheses of difficulty as explained in chapter 4 and
the empirical experiment of difficulty conducted with this
group as explained in chapter 5 in order to provide the
grammatical information that is appropriate for their needs.

6.1.5. The Rhetoric Section. When the user clicks on the
gn§§9;12_button, sentences that mark the ki 'beginning', shoo
'development', ten 'change', and ketsu 'conclusion' are
underlined. When these sentences are selected, explanations
on the rhetorical structure are provided. Figure 7 contains
the text for page 2 of ”Aruke Orinpikku". In this example,
the sentence beginning the ten 'change' in the rhetorical
structure has been selected: Ima Nihon dewa, aruki no
aikoosha ga gekizoo shi, yaku kyuuhyakuman to mo iwarete iru.
'In Japan today, those who like to walk have increased

dramatically and are said to number about 9 million.’ In the

202

Figure 7

AAAAAAAAAAAA‘AAAAAAA‘AAAAAAAAAAAAAAAAAALALAAAAAA‘LAAA‘AAAA

 

.. muoxannkkuat autumn. smite?
-: <. 127-Inmate“. fittflBJfibuflitnb. rtewm
31>] rssxv‘yfl r5151!“ catamaran“.
‘-1letamm5rtll##<. nuurstsvtrcnakvw. 9170

:ifn‘rub 0) 1M» bfl’iu.
.. $953131. «magma 11:11?th fiaettbifiwué. a—t
q-V77#-t&btalbtb.70-H-QEXQ%°T<h59§9
aub.szoztmor.eatsmiuaraunbeua.

\

IO...
U

EfflCbk%A#§<.#UVKvawflflu$!10b:$5C!vofi 1

 

.
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

 

 

 

.3 ‘ ‘ 1:2?:2:I:I:I:Z:I:Z:I:2:Z:::I:I:I:I:I:I:I;I:
i. Wu “(J a 1 \ 5 IPSfiEb T"! 1 ‘1 . lilWEt 33176 fab. it?) 244.}: .... 1:1:I:Z:I:I:I:I:1:Z:I:I:
2:: t “#73033!” 151K612 b. 1&3 (Difia) 7:3). 35 3 31:5 3&1; 3 : 3! 12:31:22 3 ::I:I:I:I:I:I:I:I:I;I:I:
2‘: f5”. 25 8 (Dam: 11511 U . 25' ( C t ‘3 § L U o: t 755 7 . 2:323: T I:I:I:I;I:I:I:I;I:I:I;I;
I 1 WW0 t:::.:::::.:.:..2.22:::2:::I:::::I:I:I:I:I:
2. ““““““““““““““““““““““““““““““ my .. [25232322252225
3 ‘ ‘ ‘ -:-:;:;:;:;z;z;2:2;2:2;:;:;:§:§:§:§:§:§:§:§:
371% ‘01 ten 5553553353533323333353;335332335333533333335

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

OOOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO

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

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

OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

 

 

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

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

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

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

203

OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOOO

window below the text, a message describes the rhetorical
function of this sentence, which is to bring about a change
in the topic. Here the topic moves from the 'Walking
Olympics' in Holland to the popularity of recreational
walking in Japan. The information provided to the users of
this program should help to make the passage more coherent by
helping them follow the progression of the passage.
According to Hinds (1983a, pp. 190-191), to native speakers
of English, a composition that exhibits the ki-shoo-ten-ketsu
pattern appears to lack coherence. Explanations about the
points of transitions in Japanese expository texts can
certainly increase the sense of coherence for English-
speaking readers. In summary, the purpose of rhetorical
explanations is to assist the readers in understanding
Japanese text by helping them to understand the rhetorical
structure contained in it.

6.1.6. The Culture Section. When the user clicks the
ggltg;g_button, sentences or sections that present cultural
difficulties are underlined. When these sections are
selected by the user, cultural explanations are given based
upon the predictions of cultural difficulty given in chapter
4 under feature 20 (section 4.1.20). Figure 8 contains the
text for page 1 of "Aruke Orinpikku". In this example, the
sentences explaining the interaction between the elderly

woman and Mr. Kaneko about the meaning of the word faito

204

Figure 8

AAAAAAAAAAAAAAAAAAAAA‘AAAAAAAA‘AAAAAA‘AAAAAAAAAAAAAAAAAAAAA

 

iii :11: arr/aw

.;: <. sumo. m mm»
5;: :wsiem rams.

IBI’a‘lfitfib f:b®1€7‘*‘c#®flf£ )1 57

353(awummeauzu:mummy“. mammals. me:
3:41: r774 r. 774 n a filulflkt‘tfipfc.

 

 

 

 

 

g 5:.zsyyowzunr.xsmmtvfimbnb.7z—?x-
5; v-ftb nun. 251131'J7897tb sun-cub. _
:3 2mm“ 1!!! filoAUtif COJII+4X-’\‘/1:!!o‘(. as ;

150mm: awer+mmm bk. em neon =1: -a/.. E:

 

OOOOOOOOOOOOOOOOOOOOOO

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

............
............
............
............
............
OOOOOOOOOOOO
OOOOOOOOOOOO
............
............

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

OOOOOOOOOOOOOOOOO

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

 

 

 

. 76¢. aeotemaasruok. !EE§§E‘§(\=»®, -

*1 ‘ v " ' “ 1‘ a[214
fifiW
:23 Thfil'lyldymbudth'oll'77‘fb'il
gfism Tbvod'774 F'Mmhfiuam

{mum h-flyf‘fio-bfiwmmbb
Z-ipfiip‘lhhvdk. hmitis-ot-dformof
Iflr—uorvmurhymycuh'od'77‘f F'vfib
I-lr‘uuvdfiuhpflow [dob-(HI
333hdbc-bM1ab-offu.

 

 

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

OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOO
Q.

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

...........
OOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOO

 

 

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
..............................................................
.............................................................

205

OOOOOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOOOO

'fight' have been selected: "Naze tatakaou to in no.
Watakushitachi wa tatakai wa gomen da wa." Ganbarou to
iikikaseru karui chooshi no "faito" datta ga, Kaneko san wa
"we don't want to fight" to in kotoba ni komerareta mono o
sasshite, fukaku atama o sageta. '“Why do you talk about
fighting. We don't want to fight.” It was a light "Fight!”
calling on people to exert an effort, but Kaneko realized the
thinking behind the words, "we don't want to fight," and hung
his head.’ In the window below the text, a message discusses
the meaning of the word faito 'fight'. The elderly lady did
not understand that when the Japanese used the word faito
during the walk, they were not using the word in its literal
sense. They were merely giving encouragement to each other.
Mr. Kaneko felt embarrassed that the use of this word had
created such a misunderstanding. The users of this program
need to know that in Japanese culture, the use of the word
faito while walking or running is not to be taken literally.
In summary, providing cultural background to a passage
written in another language facilitates the understanding of
that passage because it can provide access to the knowledge,
assumptions, and beliefs made by the people in that culture.

6.1.7. The Tests. After each passage, there is a set of six
comprehension questions based on the content of the passages.
These questions were developed by the researcher. These

questions were constructed to measure the extent of reading

206

Figure 9

....................................................................................
OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
....................................................................................
OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO
....................................................................................
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
nnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn
OOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOO

2522222222; Question 1. ....._ ,2is§2§s§2§2§2§2§3§2§s§3§2§
e 301119113. 1:1an 1: are 1’ v .
0 ”fl“-
o
o it 1; .1 6 11a; 5 1111\in a 1 ., ‘

 

 

207

comprehension on the part of the subjects. These tests
employ a multiple-choice format. Each test item employs a
question followed by a set of answers. One of these answers
is correct, and three are distractors. See Figure 9 for an
example question.
6.1.8. Monitoring. The computer is used to monitor what
kind of information the reader uses while working through the
lesson. The program records the number of times information
on vocabulary, Chinese characters, grammar, rhetoric, and
culture is used in order to understand a text. An analysis
of this information over a group of users can determine the
difficulty of a text. The assumption is that the more
difficult a certain element of text is, the more the users
need information to explain that element. There is a
proportional relationship between the number of times
information about a textual element is accessed and the
difficulty of that element. Difficulty is determined by the
number of times information on certain textual elements is
used. This sort of information is used to analyze the
sources of textual difficulty, and thus, the computer can be
an aid in linguistic study.

This program keeps a record of both individual user
performances and the group performance as a whole. The
number of times and relative percentages of help received in

vocabulary, kanji, grammar, rhetoric, and culture are

208

recorded on an individual and group basis. Thus, a record of
the total number of times information is used on an
individual and group basis is kept. Also, there is a record
of the length of time it took the readers to read the
passages. Also, the program keeps a tabulation of the help
received according to the twenty experimental features.
Furthermore, a record is kept of each student's answers and
scores while taking the two comprehension tests. In
addition, there is a record of the time it took to complete
the tests. The computer also tallies a complete summary of
the scores and times on a group basis. This information is
used to evaluate textual difficulty according to the
assumption that the more difficult a certain element of text
is, the more the readers need information to explain that
element. In addition, the test scores can be used to compare
computer-mediated text with text that is presented in an off-
line condition. With this data, empirically-based
conclusions can be drawn on whether the computer can
facilitate the reading of Japanese as a second language on
the part English-speaking readers.

6.1.9. The Procedure. The passages used in this computer
program were the same as those that were used in the
experiment in chapter 5. The subjects for the experiment in
chapter 5 were students at the end of the third year of

Japanese. This same group of students was used for the

209

computer experiment, this time at the beginning of the fourth
year of Japanese. In order to avoid the practice effects
involved in giving the texts to the whole group, only half of
the third-year students were involved in the experiment in
chapter 5 This way only some of the students in the subject
group for the computer experiment had seen the texts before.
This is important because only some of the students were
influenced by the practice effects of seeing the texts
before.

The subjects were assigned to two separate experimental
groups: 1) the group that read the passages in the off-line
condition on printed pages, and 2) the group that read the
passages in the on-line condition on the computer monitor.
The subjects in group 1 were given as much time as necessary
to read the passages. The amount of time it took to read the
passages was recorded. When they finished reading one
passage, they were supplied with the six printed
comprehension questions that went with that passage (see
Appendix C). The time it took to take the test was recorded.
After the test was finished, the second text was given out.
When the second text was finished and the time it took to
read the text was recorded, they took the second test (see
Appendix C). When they finished the test, the time it took
to take the second test was recorded. At this point, they

were finished.

210

The subjects in group 2 were given an introduction to
how a computer operates before they read the passages. The
computer program itself contained an introduction to how the
program worked. Before even reading the passages, the
subjects had to work through this introduction. Once the
subjects finished the introduction, they read the computer-
mediated text version of the first text, "Hoogen to
Kyootsuugo", using any of the help facilities they needed to
understand the passage. Once they finished the passage, they
took the comprehension test based on the content of the text
(see Appendix C). This was the same test as used in the off-
line condition except that the subjects used the mouse to
select the answers on the monitor. Once the test was
finished, the subjects read the second text, ”Aruke
Orinpikku". When that was finished, they took the second
test (see Appendix C). When the second test was completed,
the program came to an end. The time it took to read the
passages and do the tests was automatically recorded by the
computer. The computer also recorded all the help the
subjects received and the scores they got on the tests.

6.2. The Results. This section presents a summary of the
data. Further expanded discussion of the implications of

this data is presented in chapter 7.

211

6.2.1. On—line Help Received. In section 6.1.8. of this
chapter, an assumption was put forth. It stated that the
more difficult a certain element of text is, the more the
reader needs information to explain that element. This
indicates that there is a proportional relationship between
the number of times information about a textual element is
accessed in a computer program and the difficulty of that
element. Difficulty is determined by the number of times
information on certain textual elements is used. Using this
sort of information, the passages can be analyzed to
determine the sources of textual difficulty.

The following chart gives the percentages of the number

of times the users requested help while reading these

passages:
Table 9

A£Q£§.QI1£21&KQ .fl929§£_22_5222£§2292
Grammar 0% 8.70%
Vocabulary 61.9% 82.61%
Kanji 35.71% 8.70%
Rhetoric 0% 0%
Culture 2.38% 0%

6.2.2. Test Scores. Each passage was followed by six
multiple-choice questions containing four choices with one of

the choices constituting the correct answer. The students

212

who read the passages in the on-line condition and those who
read the passages in the off-line condition took the same
tests. The average test scores for both the on-line and off-
line groups are indicated in the following table. The
averages are calculated based on 6 possible points for each
test (one point for each question).

Table 10

W W
On-line 0.5 2.5

6.2.3. Passage Reading Time. The length of time to read the
passages in both the on-line and off-line condition is

compared in the following chart:

Table 11
WW
On-line 11 mins. 5.5 secs. 7 mins. 56 secs.
Off-line 9 mins. 28.5 secs. 6 mins. 28 secs.

6.3. Discussion. This section presents some initial
discussion. Further discussion of the implications of this
data is presented in the next chapter. The biggest surprise
is the fact that information on grammar was accessed very

little by the students. Vocabulary was certainly accessed

213

the most. This finding about the importance of vocabulary
for readers is supported by the research as reported in the
literature (Anderson and Freebody, 1981; Coleman, 1971;
Freebody and Anderson, 1981; Klare, 1974-1975; Lado, 1972;
Marks, Doctorow, and Wittrock, 1974; Thorndike, 1973), which
states that vocabulary is a strong predictor of overall
textual difficulty. Another surprising finding is that the
students reading the passages in the off-line condition had a
better average on both tests. Possible reasons for this are
discussed in chapter 7. Not surprisingly, another finding is
that students reading the passages in the on-line condition
were slower. It took time for them to access all the on-line
help in vocabulary, kanji, grammar, rhetoric, and culture.
This tended to slow them down. The next chapter further

discusses the implications for this data.

214

CHAPTER 7

Discussion

7.0. Introduction. As stated in Chapter 2, the main purpose
of this pilot study was to examine the effect of syntactic
variables on L2 reading of Japanese. This study addressed
the following two hypotheses:
1. Grammatical variables affect the difficulty of the
reading of a Japanese text by native speakers of

English.

2. Computer-mediated text can facilitate the
comprehension of a passage more than ordinary text.

In testing the first hypothesis, this pilot study focused on
the contributions of grammatical structure to passage
difficulty. As a part of this study, the contributions of
rhetorical structure, vocabulary, and culture to textual
difficulty were also considered in order to complete the
total picture of the major factors that affect the level of
difficulty of Japanese text. All these factors help
contribute to the difficulty that faces the reader of
Japanese as a second language. This pilot study considered a
list of 20 features which can contribute to passage
difficulty. This list is an enumeration of 20 possible
variables that are hypothesized to influence passage
difficulty. This is not an exhaustive list, but a list that

can provide some insight into textual difficulty. These

215

variables are mainly syntactic in nature which is the focus
of this study; however, variables that are rhetorical,
lexical, and cultural in nature were also considered in order
to provide a more balanced study of the sources of textual
difficulty. In order to test the second hypothesis, computer
software was designed with the 20 variables in mind in order
to determine 1) whether it can measure quantitatively the
contribution these factors make to passage difficulty and 2)
whether it can actually facilitate the reading of Japanese
text. This software was tested on human subjects in order to
measure its effectiveness in measuring textual difficulty and
its power to facilitate the comprehension of Japanese written
text.
7.1. The Study. Two Japanese texts were evaluated in light
of grammatical, rhetorical, lexical, and cultural issues
affecting textual difficulty. Most of the features used in
the evaluation were grammatical in nature, so the
constructions in these passages which contributed to
grammatical complexity were focused on in order to illustrate
some of the issues involved in passage difficulty. Each one
of the features was a specific hypothesis as to what
contributes to passage difficulty in reading comprehension.

In order to empirically examine these features, two
Japanese texts, one rated difficult and the other easy with

respect to these 20 features, were investigated as to the

216

actual difficulty they presented to English-speaking students
who were reading Japanese as a second language. The readers
were asked to read these texts and circle or underline the
parts of the texts they thought were difficult. The purpose
of this experiment was to critique the validity of the
assessment of difficulty according to the 20 features. The
actual analysis of the predictive ability of the 20 features
is based on two calculations: 1) the percentage of
predictions that were correct and 2) the actual problem areas
as determined by this experiment that were successfully
predicted. These calculations were explained in chapter 5.
Based upon both the linguistic analysis (as contained in
chapter 4) and the results of the experiment assessing the
difficulty in both texts (as contained in chapter 5), a
computer program was written focusing on the potential
problem areas in the reading comprehension of these two
passages. The program focused on the areas of difficulty
that were indicated by this linguistic analysis (chapter 4)
and the evaluation of the performance of the target
population during the experiment assessing the difficulty in
both texts (chapter 5). The computer program provided help
in these areas of anticipated difficulty for the target
population. The goal was to provide the most appropriate

help to intermediate Japanese students.

217

In this computer program, computer-mediated text
was manipulated in a variety of ways. The reader interacted
with the passage which was placed on the computer screen in a
text window. At any point during the reading process, the
reader had certain resources available which could be used to
facilitate understanding of the text. These resources were
in the form of a help menu which presented certain choices to
facilitate comprehension.

The computer was used to monitor the kind of information
the reader used while doing the lesson. It kept track of the
number of times information on vocabulary, Chinese
characters, grammar, rhetoric, and culture was accessed in
order to understand the text. An analysis was made of this
information over a group of users to determine what was
difficult in text. The assumption was that the more
difficult a certain element of text is, the more the users
need information to explain that element. There was a
proportional relationship between the number of times
information about a textual element is accessed and the
difficulty of that element. As a result, difficulty was
determined by the number of times information on certain
textual elements was accessed. This sort of information was
used to acquire an overall picture of the sources of textual

difficulty.

218

The software developed for this pilot study was
evaluated under two treatment conditions. The first
treatment condition presented text in an off-line condition
in conventional print on a piece of paper, and the second
presented text in an on-line condition on the computer screen
with options on a help menu to aid in comprehension.
Students read the two passages under one of these two
treatment conditions. One group of students read the texts
in the off-line treatment condition (condition 1) and the
other group in the on—line treatment condition (condition 2).
Following the passages in both treatment conditions, the
students were presented with a series of six multiple-choice
comprehension questions giving four options with three
distractors and one correct answer. While not looking at the
passages, the students in both treatment groups answered
these questions as a test of their reading comprehension.
Besides keeping track of the comprehension score, the program
kept track of the time it took to read the passages. Also, a
record was kept of the amount of time needed to complete the
comprehension questions. In addition to this, a record was
kept of the help the subjects in the on-line treatment group
used in order to facilitate understanding of the text. In
summary, this pilot study was carried out to investigate the

causes of difficulty in Japanese texts and the possible

219

contributions computer-mediated text can. make 1x) the
comprehension of text.
7.2. The Results.
7.2.1. The Implications of the Data of Chapter 5.
7.2.1.1. The Implications for ”Aruke Orinpikku". A major
difference in the predicted problems for this passage, as
given in chapter 4, and the actual problems, as encountered
in chapter 5, was the relative percentage of the problems
contributed by grammar, vocabulary, kanji, rhetoric, and
culture to the total number of problems. Table 12 compares
these predictions (in chapter 4) with the actual problems (in
chapter 5):

Table 12

WW

Grammar 42.71% 15.83%
Vocabulary 24.62% 42.45%
Kanji 31.16% 41.73%
Rhetoric 1.01% 0%
Culture 0.50% 0%

.According to the predictions of chapter 4 made on the basis
of the 20 features, grammar at 42.71% of the total number of
problems was expected to have the most influence on the
difficulty of this passage. However, in the experiment in

chapter 5, which assessed the actual difficulty of this

220

passage, vocabulary at 42.45% and kanji at 41.73% had the
most impact on difficulty while grammar was at only 15.83%.
The findings in this experiment about the importance of
vocabulary are supported by the reported research in the
literature (Anderson and Freebody, 1981; Coleman, 1971;
Freebody and Anderson, 1981; Klare, 1974-1975; Lado, 1972;
Marks, Doctorow, and Wittrock, 1974; Thorndike, 1973), which
states that vocabulary is a strong predictor of overall
textual. difficulty. In considering overall textual
difficulty, vocabulary difficulty and kanji difficulty are
important considerations, but they need to be balanced
against other factors, such as linguistic structure.

Comparing the prediction success rate with the actual
success rate in relation to the areas of grammar, vocabulary,
kanji, rhetoric, and culture provides further evaluation of
the predictions made in chapter 4. Table 13 gives the
prediction success rate and the actual success rate for each
of the above areas:

Table 13

EI§9i_§QQQ§§§_BQ§§. AEEB§1_§BQQQ§§_BQEB.

Grammar 24.71% 90.91%
vecabulary 77.08% 62.71%
Kanji 62.90% 67.24%
Rhetoric 0.00% 0.00%
Culture 0.00% 0.00%

221

For grammar, the prediction success rate was quite low at
24.71% while the actual success rate was quite high at
90.91%. The reason for this was that too much emphasis was
placed on grammar in making predictions in chapter 4. The
problems the students actually had with grammar were
predicted successfully, but in the process too many extra
grammatical predictions were made. The students seemed to be
preoccupied with problems involving vocabulary and kanji,
rather than with problems involving grammar. This is not to
say that they do not have problems with grammar, but they
seemed to think that information on vocabulary and kanji was
more crucial for textual understanding. Thus, the
predictions for vocabulary and kanji were much more accurate
because the students needed to know vocabulary and kanji to
understand the passage. The prediction and the actual
success rates for rhetoric were 0% because no predictions for
rhetoric were fulfilled. This appears to be a limitation of
the experimental instrument of chapter 5 because it does not
appear to measure any rhetorical interference at all with
reading comprehension. Perhaps, the students themselves need
not be aware of rhetorical structure to basically understand
this passage. Also the prediction and the actual success
rates for culture were 0% because this experimental
instrument picked up no cultural interference in this

passage. Either there was no problem with cultural

222

understanding here, or this experimental instrument just
could not detect it.

A further evaluation of the predictions made in chapter
4 can be made by comparing the ranking for the grammatical
features in terms of their predicted influence on passage
difficulty (chapter 4) with the same grammatical features in
terms of the magnitude of their actual influence (chapter 5).
In Table 14, the predicted influence of these features on
passage difficulty is compared with the actual influence
(excluding features 5, 17, 18, 19, and 20). The percentages
indicate the relative percent of grammatical influence of
each feature on textual difficulty. The higher the

percentage, the greater the influence.

Table 14
WW
1. Homophonous 4.71% 0%
Particles
2. Nominalizers 7.06% 0%
3. Anaphora 1.18% 0%
4. Tense 3.53% 18.18%
6. Noun Head 8.24% 0%
Modification
7. Conjunctions 3.53% 0%
8. Subordination 18.82% 18.18%
9. Passive 10.59% 31.82%
10. Causative 2.35% 9.09%
11. Verb Complexes 4.71% 9.09%
12. Quotations 12.94% 9.09%
13. Relative Clauses 17.65% 4.55%
14. Hedges 2.35% 0%
15. Further 1.18% 0%
Embeddings
16. Comparatives 1.18% 0%

223

The actual influences as indicated in the right-hand
column are the results of the empirical study undertaken in
chapter 5. These results test the predictions in the left-
hand column that were given in chapter 4 about the influence
of grammatical features on textual difficulty. These
percentages were calculated by dividing the number of
problems in each feature category by the total number of
grammatical problems. For example, the total number of
predicted problems for the feature, Homophonous Particles,
was 4. Since the total number of predicted grammatical
problems was 85, the predicted relative percent influence of
homophonous particles on grammatical difficulty was 4.71%
(which is 4/85 x 100%). However, the actual number of
problems for the feature, Homophonous Particles, was 0
according to the experiment given in chapter 5. According to
that experiment, the total number of actual grammatical
problems was 22, so the actual relative percent influence of
homophonous particles on grammatical difficulty was 0% (which
is 0/22 x 100%), which was at variance with the predicted
amount, 4.71%.

Examining this table, it can be seen that the actual
grammatical influences are at variance with the predictions.
The closest match was subordination with the predicted
influence at 18.82% and the actual influence at 18.18%.

Subordination was hypothesized to be an important source of

224

grammatical difficulty, and the empirical study of chapter 5
confirmed that it was. This is because subordination has to
do with the embedding of clauses, which adds to sentence and
therefore passage difficulty. Relative clauses were
predicted to be a major source of grammatical difficulty at
17.65%, but the empirical study indicated that they were not
as great a problem because the percentage was 4.55%.
Quotations were predicted to be another major contribution to
grammatical difficulty at 12.94%, and the actual contribution
was 9.09% which was reasonably close to the prediction.
Quotations increased the embedding, thus adding to sentence
complexity. The combined feature categories {of
subordination, quotations, and relative clauses indicate that
clause embedding was 31.82% (which is the sum of 18.18%,
9.09%, and 4.55%) of the actual grammatical problems and an
important component of passage difficulty. 7

Verb morphology was predicted to be 16.47% of the
grammatical difficulty, which is the sum of passive, 10.59%,
tense, 3.53%, and causative, 2.35%. However, surprisingly,
verb morphology was 59.09% of the 321.113.]. grammatical
difficulty, the sum of passive, 31.82%, tense, 18.18%, and
causative, 9.09%. Verb morphology was difficult for the
students partly because they needed to understand the case
roles involved with the passive and causative. Perhaps, the

students also had trouble with the Japanese tense system.

225

Verb complexes also gave the students trouble at 9.09%; the
prediction was 4.71%. The students had trouble with the
compounding of verbs. This difficulty is mainly related to
the semantics involved in combining two verb forms together.

The features involving homophonous particles, noun head
modification, and nominalizers did not measure up to the
predictions of chapter 4. In reality, homophonous particles
were not a source of difficulty. The correct interpretation
of homophonous particles depends on the context in which they
are found. A clear context can eliminate the ambiguity posed
by homophonous particles, thus decreasing the difficulty
posed by them. This was the case in this passage. Noun head
modification had an influence on difficulty according to the
empirical experiment in chapter 5 but not quite as much as
predicted in chapter 4. Modification of nouns by other nouns
and adjectives did not seem to be a major problem. The
actual influence for nominalizers was 0% although it was
predicted to be 7.06%. The nominalizers themselves were not
a problem, but the embedding they produced perhaps caused
difficulty.

The rest of the features had a 0% actual influence, but
they were not predicted to have much influence anyway, so
that is not surprising. It would be difficult for any one

single passage to contribute to difficulty in all of the

226

twenty feature categories. Each passage presents its own
unique difficulties and challenges to the reader.

7.2.1.2. The Implications for ”Hoogen to Kyootsuugo". As
with "Aruke Orinpikku", there were differences in the
predicted problems for this passage, as given in chapter 4,
and the actual problems, as encountered in chapter 5. Table
15 compares these predictions (in chapter 4) with the actual

problems (in chapter 5):

Table 15
WW
Grammar 61.36% 19.30%
Vocabulary 25.00% 50.88%
Kanji 9.09% 26.32%
Rhetoric 2.27% 0%
Culture 2.27% 3.51%

According to the predictions of chapter 4, grammar at 61.36%
of the total number of problems was predicted to influence
passage difficulty most, but in the experiment in chapter 5,
vocabulary at 50.88% had the most impact on difficulty while
grammar was at only 19.30%. These findings supporting the
importance of vocabulary are backed up by the research in the
literature as cited above. However, as mentioned above, in
considering overall textual difficulty, although vocabulary

difficulty is an important consideration, it needs to be

227

balanced against other factors, such as linguistic structure.
The prediction for the kanji was 9.09%, but the actual impact
on difficulty was 26.32%. This means that in actuality kanji
is second to vocabulary in determining passage difficulty.
The prediction for culture at 2.27% was close to the actual
difficulty at 3.51%. The prediction for culture was fairly
easy because it involved Japanese place names which could be
circled by the students doing the experiment if they felt
that these place names were difficult to understand. The
method works best at a strictly local level where one can
circle the problem.

When the prediction success rate is compared with the
actual success rate in relation to the areas of grammar,
vocabulary, kanji, rhetoric, and culture, the predictions
made in chapter 4 can be further evaluated. Table 16 gives
the prediction success rate and the actual success rate for

each of the above areas:

Table 16
Ereglaflmzzeuijsnaz Aefnal_§ug§e§§_fiane
Grammar 18.52% 100%
Vocabulary 63.64% 51.72%
Kanji 37.50% 20.00%
Rhetoric 0.00% 0.00%
Culture 100% 100%

228

The prediction success rate for grammar was quite low at
18.52% while the actual success rate was 100%. As stated
above, the reason for this was that too much emphasis was
placed on making predictions for grammar in chapter 4. The
problems the students really had with grammar were predicted
successfully, but in the process too many extra grammatical
predictions were made. The students seemed to focus on
problems involving vocabulary and kanji, rather than grammar.
To these students, information on vocabulary and Chinese
characters was more important for the comprehension of
Japanese text. Thus, the predictions for vocabulary and
kanji were more accurate because the students needed to know
vocabulary and kanji to understand the passage. As a result,
they marked more vocabulary and kanji difficulties. As with
"Aruke Orinpikku", the prediction and the actual success
rates for rhetoric were 0% because no predictions for
rhetoric were fulfilled. Again, the experimental instrument
of chapter 5 does not appear to measure any rhetorical
interference at all with reading comprehension. In addition,
the prediction and the actual success rates for culture were
100% because the students only circled the predicted cultural
problems. The prediction for culture was not difficult
because the students could easily circle the Japanese place

names that were hard to understand.

229

As with the passage, "Aruke Orinpikku", a further
evaluation of the predictions made in chapter 4 for "Hoogen
to Kyootsuugo" can be made by comparing the ranking for the
grammatical features in terms of their predicted influence on
passage difficulty (chapter 4) with the same grammatical
features in terms of the magnitude of their actual influence
(chapter 5). In Table 17, the predicted influence of these
features on passage difficulty is compared with the actual

influence (excluding features 5, 17, 18, 19, and 20):

Table 17
WW
1. Homophonous 7.41% 0%
Particles
2. Nominalizers 7.41% 0%
3. Anaphora 1.85% 0%
4. Tense 3.70% 0%
6. Noun Head 18.52% 0%
Modification
7. Conjunctions 7.41% 9.09%
8. Subordination 18.52% 18.18%
9. Passive 7.41% 36.36%
10. Causative 0% 0%
11. Verb Complexes 1.85% 9.09%
12. Quotations 5.56% 0%
13. Relative Clauses 20.37% 27.27%
14. Hedges 0% 0%
15. Further 0% 0%
Embeddings
l6. Comparatives 0% 0%

The percentages printed in the chart give the relative
percent of grammatical influence of each feature on textual
difficulty. The higher the percentage, the greater the

grammatical influence of that feature. The purpose of this

230

table is to show the grammatical features that have the most
influence on passage difficulty. Given this information, it
is possible to have a more accurate idea of the features that
influence difficulty the most.

As with table 14, the actual influences as indicated in
the right-hand column are the results of the empirical study
undertaken in chapter 5. These results test the predictions
in the left-hand column that were made in chapter 4 about the
influence of grammatical features on textual difficulty.
These percentages were the results of dividing the number of
problems in each feature category by the total number of
grammatical problems. To give an example, the total number
of predicted problems for the feature, Conjunctions, was 4
for "Hoogen to Kyootsuugo". Since the total number of
predicted grammatical problems was 54, the predicted relative
percent influence of sentence-level conjunctions {on
grammatical difficulty was 7.41% (which is 4/54 x 100%).
However, the actual number of problems for the feature,
Conjunctions, was 1 for this passage according to the
experiment given in chapter 5. According to that experiment,
the total number of actual grammatical problems was 11, so
the actual relative percent influence of sentence-level
conjunctions on grammatical difficulty was 9.09% (which is

1/11 x 100%), close to the predicted percentage.

231

Examining this table, the actual grammatical influences
for ”Hoogen to Kyootsuugo" are at various degrees of variance
with the predictions. As with "Aruke Orinpikku", the closest
match involved subordination with the predicted influence at
18.52% and the actual influence at 18.18%. Subordination was
hypothesized to be an important predictor of grammatical
difficulty, and the empirical study of chapter 5 confirmed
this hypothesis. Subordination involves the embedding of
clauses, contributing to sentence and passage difficulty.
Relative clauses were predicted to be a major source of
grammatical difficulty at 20.37%, and the empirical study in
chapter 5 showed that they were because the actual
contribution to grammatical difficulty was 27.27%. In
”Hoogen to Kyootsuugo", branching direction seems to be an
important issue in comprehending relative clauses, a fact
which supports the predictions. Quotations were predicted to
make a modest contribution to grammatical difficulty at
5.56%, but the actual contribution in this passage was 0%,
which indicates they ‘weree not significant source of
difficulty in this passage. The combined feature categories
of subordination and relative clauses indicate that clause
embedding was 45.45% (which is the sum of 18.18% and 27.27%)
of the actual grammatical problems and a very important
component of passage difficulty. This degree of embedding is

reflected in the average sentence length for this passage,

232

which is 21.27 words, compared to an average sentence length
of 9.97 for "Aruke Orinpikku", which had a lower embedding
percentage of 31.82%.

Verb morphology in "HOogen to Kyootsuugo" was predicted
to be 11.11% of the grammatical difficulty, which is the sum
of passive, 7.41%, and tense, 3.70%. However, surprisingly,
verb morphology was 36.36% of the ml grammatical
difficulty, the sum of passive, 36.36%, and tense, 0%. The
passive was difficult for the students because they needed to
understand the positioning of the case roles involved with
this voice. In the passive, the agent (if it appears) is
moved from the subject position to an unfocused adjunct
position, and the undergoer takes its place. The student
needs to be aware of this change in the functional
positioning of these roles. Verb complexes also gave the
students trouble at 9.09%. The students‘ trouble with the
verb compounding was mainly related to the semantics involved
in combining two verb forms.

Homophonous particles and noun head modification had no
influence on difficulty in this passage according to the
empirical experiment in chapter 5 although a fair amount of
influence had been predicted in chapter 4 (homophonous
particles at 7.41% and noun head modification at 18.52%). In
”Aruke Orinpikku” (see section 7.2.1.1), homophonous

particles and noun head modification had some influence on

233

passage difficult but not quite as much as predicted.
Obviously, these two features did not contribute as much to
textual difficulty as predicted in this experiment. As
mentioned is section 7.2.1.1 in the discussion of "Aruke
Orinpikku”, the correct interpretation of homophonous
particles depends on the context in which they are found. A
clear context can eliminate the ambiguity posed by
homophonous particles, thus decreasing the difficulty posed
by them. This is the case also in this passage, ”Hoogen to
Kyootsuugo". As with "Aruke Orinpikku", noun head
modification did not seem to be a major source of difficulty
in this passage for the students.

Also, in ”HOogen to Kyootsuugo”, the actual influence
for nominalizers was 0% although it was predicted to be
7.41%. This is the same result (0%) as that found in ”Aruke
Orinpikku", so nominalizers do not seen to be a major
problem. It seems the nominalizers themselves were not a
problem, but the embedding they produced caused the
difficulty.

The rest of the features have a 0% actual influence, but
they were predicted to have little or no influence anyway, so
that is expected. It would be difficult for any one single
passage to contribute to difficulty in all of the twenty
feature categories. Each passage contains its own unique set

of difficulties.

234

7.2.1.3. Summary of the Implications. In examining the
results of the experiment on the sources of difficulty in
"Aruba Orinpikku” and ”Hoogen to Kyootsuugo" in chapter 5,
vocabulary has been shown to be the most significance factor
in determining textual difficulty. This is in accord with
previous research in the literature on reading theory.
Chinese characters are also an important factor in textual
difficulty for readers of Japanese as a second language.
Grammar is another important factor in textual difficulty
although not as important as vocabulary. Two major factors
involved in grammatical difficulty stand out in the
experiment: embedding and verb morphology. The embedding
apparent in these passages involves subordination,
quotations, and relative clauses. This embedding adds
complexity and length to sentences. Verb morphology involves
verb tense, the passive, and the causative. Verb morphology
was difficult for the students because they needed to
understand the Japanese tense system and also the case roles
involved with the passive and causative. Rhetorical
structure and cultural background also contribute to textual
difficulty, but it is an effect that is difficult to measure
by the empirical instrument in chapter 5. Further research
needs to investigate in more detail the effect of rhetorical

structure and culture on difficulty in text.

235

The study in chapter 5 also confirmed that "Aruke
Orinpikku" (predicted difficulty index = 0.64) is more
difficult than "Hoogen to Kyootsuugo" (predicted difficulty
index = 0.47) as predicted in chapter 4, section 4.2. The
actual difficulty for these passages, based on the experiment
in chapter 5, confirms these predictions. The actual
difficulty index of a passage is the total number of actual
problems in a passage divided by the number of words in that
passage. For each passage, this index can be further divided
into sub-indices for grammar, vocabulary, kanji, rhetoric,
and culture. These are calculated by dividing the number of
actual problems in each of these sub-areas by the total
number of words in the passage under consideration. Table 18
contains the results of these calculations for actual

difficulty, based on the experiment in chapter 5:

Table 18
WW
Grammar 0.07 0.05
Vocabulary 0.18 0.12
Kanji 0.18 0.06
Rhetoric 0.00 0.00
Culture 0.00 0.01
Total Difficulty 0.42 0.24

236

According to these calculations, "Aruke Orinpikku"
(difficulty index = 0.42) is more difficult than "Hoogen to
Kyootsuugo" (difficulty index = 0.24) confirming some of the
predictions. All the sub-indices of "Aruke Orinpikku" are
larger than those of "Hoogen to Kyootsuugo" except for
culture. However, the difference in the culture indices was
minimal. The interesting thing is that the index of
difficulty for grammar is larger for "Aruke Orinpikku" than
for "Hoogen to Kyootsuugo" even though the average sentence
length for ”Hoogen to Kyootsuugo” (21.27) is longer than for
”Aruke Orinpikku" (9.97). This indicates that even if the
average sentence length in one passage is longer than in the
other, it does not guarantee that the passage with the longer
average sentence length is more difficult grammatically.
This is interesting because sentence length is an important
variable in readability formulas, which estimate passage
difficulty. These formulas are too superficial. This
research indicates that what is needed is a deeper analysis
of grammatical structure in order to determine grammatical
difficulty and refine the formulas.

7.2.1.4. Conclusions. Based upon the empirical study of
chapter 5, syntactic variables, such as tense (4),
subordination (8), passive (9), causative (10), quotations
(12), and relative clauses (13), have been shown to cause

difficulty in the reading of Japanese text by native speakers

237

of English. In section 2.6 of chapter 2, it was stated that
there is a gap in the reading research of Japanese as a
second language relating to the effect that textual variables
have on the readability of Japanese text. This study was a
first attempt to fill this gap. Further research needed to
be done, so an additional step in this research project was
to use the computer to investigate the effect of syntactic
variables on textual difficulty. In addition to this, this
project investigated the role of the computer in facilitating
the reading of Japanese text. In the next section of this
chapter, the results of this further study will be discussed.

7.2.2. The Implications of the Data of Chapter 6.

7.2.2.1. The Subjects. The computer experiment was
conducted on 6 fourth-year Japanese students. In the one-
hour time period allotted for the experiment, only 2 students
completed the computer experiment (one—half hour for each
student) while four students completed the test in the off-
line condition. The experiment in the on-line condition took
longer for each student than did the experiment in the off-
line condition. Also, there was only one computer, so only
one student could use the computer at a time. The students
reading the passages in the off-line condition could work
simultaneously since the passages and tests were printed on
ordinary pieces of paper. As a result, more of them could do

the test in the allotted time period. The subjects

238

themselves were not separated according to their ability.
The subjects reading the passages in the on-line condition
had seen the passages before during the test given in the
spring of 1991; however, that experience did not seem to give
them an advantage. As a result, the practice effects caused
by seeing the passages before can apparently be ignored. The
fact that the number of students that participated in this
experiment was rather small lead to some unpredictable
results.

7.2.2.2. The Contributions to Passage Difficulty. The
contributions of grammar, vocabulary, kanji, rhetoric, and
culture to textual difficulty as predicted in chapter 4, and
determined empirically by the experiments in chapters 5 and 6

for “Aruke Orinpikku" are given in table 19:

Table 19

mm mm

(Chapter 4) (Chapter 5) (Chapter 6)
Grammar 42.71% 15.83% 0%
Vocabulary 24.62% 42.45% 61.90%
Kanji 31.16% 41.73% 35.71%
Rhetoric 1.01% 0% 0%
Culture 0.50% 0% 2.38%

According to the predictions in chapter 4, grammar at 42.71%
was predicted to have the most influence on passage

difficulty. However, according to the results of the

239

experiments in chapters 5 and 6, grammar is not seen as a
major component in passage difficulty. The fact that grammar
had a zero percentage in the computer experiment for this
passage could have been caused by the fact that only two
students did the on-line experiment. According to these
experiments, vocabulary has an important impact on passage
difficulty. This is in accord with the findings as published
in the literature as mentioned previously. According to the
predictions in chapter 4, kanji at 31.16% was predicted to
have an important impact on passage difficulty. This was
born out by both experiments. Also rhetoric and culture were
predicted not to have a major effect on passage difficulty.
This prediction was supported by these experiments. In
considering overall textual difficulty, although vocabulary
and kanji are the most important considerations, linguistic
structure, rhetoric, and culture do need to be considered in
order to create a balanced approach to textual understanding.
The contributions of grammar, vocabulary, kanji,
rhetoric, and culture to textual difficulty as predicted in
chapter 4, and determined empirically by the experiments in
chapters 5 and 6 for "Hoogen to Kyootsuugo" are given in

table 20:

240

Table 20

mm mm:

(Chapter 4) (Chapter 5) (Chapter 6)
Grammar 61.36% 19.30% 8.70%
vocabulary 25.00% 50.88% 82.61%
Kanji 9.09% 26.32% 8.70%
Rhetoric 2.27% 0% 0%
Culture 2.27% 3.51% 0%

Here grammar fares a little better in the computer experiment
at 8.70%. However, grammar is not the influence on textual
difficulty it was predicted to be. The grammatical features
of noun head modification (feature 6), passive (feature 9),
and relative clauses (feature 13) received 1 point each in
the computer experiment. This means information on these
feature categories was requested only once. Apparently, the
two students did not think grammar was a major problem. It
is interesting to note that the grammatical information that
was requested did involve the passive construction and clause
embedding. This corroborates the results of the experiment
in chapter 5, discussed in section 7.2.1 of this chapter,
which analyzed the importance of the contribution of verb
constructions and embedding to textual difficulty. According
to the experiments in chapters 5 and 6, vocabulary is the
most important consideration in textual difficulty for this

passage, in accordance with the findings of current research.

241

The experimental results for kanji (8.70%) in the computer
experiment correspond closely to the prediction (9.09%).
Chinese characters were not used a great deal in this
passage, so as a result, they were not a major consideration.
As in "Aruke Orinpikku", rhetoric and culture were predicted
not to have a major effect on difficulty in this passage.
This prediction was supported by these experiments. In this
passage also, when considering overall textual difficulty,
vocabulary needs to be considered along with linguistic
structure, kanji, rhetoric, and culture.

It is interesting to note that the students requested
information from the computer 65 times. For the passage,
"Hoogen to Kyootsuugo", 23 requests for information were made
while 42 requests were made for "Aruke Orinpikku". This
suggests that ”Aruke Orinpikku” was the more difficult
passage since the students needed more help to understand it.
This result coincides with the prediction made in chapter 4
that "Aruke Orinpikku" was the more difficult passage.
7.2.2.3. Discussion of the Test Scores. One of the most
surprising results of the computer experiment was that the
off-line group did better in the multiple-choice
comprehension tests than the on-line group did. There are
several explanations for this. One student complained that
the fonts were hard to read on the computer screen, which

would partially explain the fact that students using the

242

computer program were slower in reading the passages and
performed more poorly on the tests. Another explanation is
that the better students read the texts in the off-line
condition. A larger number of students would have perhaps
created a more balanced testing sample. Another possibility
is that the students who worked with the computer were not
computer-literate enough to feel comfortable in utilizing all
the help facilities. From their test scores, it would seem
that they under-utilized the computer. They should have
requested more grammatical information. Perhaps, they were
overconfident as to their grammatical knowledge. Perhaps,
other factors are involved. Only further research will help
to clarify the situation.

7.3. Conclusion. In conclusion, grammatical variables have
been shown to affect estimates of textual difficulty by
English-speaking readers of Japanese as a second language.
This small pilot study was a first attempt to fill a gap in
the research into reading Japanese as a second language in
relation to the effect textual variables have on the
readability of Japanese text. In any future study, the
grammatical features will have to be weighted differently
based on the results of this study. For example, important
features involving clause embedding (e.g., subordination,
quotations, and relative clauses) will have to be given a

greater weight. Other less important features (e.g.,

243

homophonous particles and comparatives) should be given less
weight. Also, based on the outcome of this study, the
variables involving vocabulary and kanji should be given more
weight than the grammatical features. In addition, the
effect of rhetoric and culture on passage difficulty needs to
be investigated further in future research since this study
did not shed much light in this area.

Furthermore, this was the first time a computer was
utilized to monitor the use of resources on the part of the
students observed in this experiment as they attempted to
understand Japanese text. However, this experiment did not
show that the computer could facilitate the comprehension of
Japanese text more than the simple reading of an ordinary
printed page. This is not the final word, however. Testing
‘with more students needs to be done. Perhaps, then the power
of the computer to facilitate the reading comprehension of
Japanese can be verified. Different kinds of experiments can
be designed that measure the reading comprehension of
students more accurately. Perhaps, instead of multiple-
choice questions, more open-ended short-answer or essay-type
questions should be asked. Only further experimentation will
help to clarify the role of the computer in facilitating the
comprehension of text for English-speaking readers of

Japanese as a second language.

244

Appendix A

Analysis of of the Japanese Texts

Aruke Orinpikku
Walking Olympics

1. Mainen oranda no chiisa na machi de ooki na lgzggphi ga

6
okonawareru.
9

‘In a small town in Holland, a big event is held every
year.‘

Grammar = 2 points
Vocabulary = 1 point
Kanji = 1 point
Length = 11 words

Foodeezu.Maachi to mo yobare, aruke orinpikku to mo
12 9 12
yobarete iru.
9

'It is called the four days' march; it is also called the
walking Olympics.‘

Grammar = 4 points
Vocabulary = 0 points
Kanji = 0 points
Length = 11 words

245

3. Nimannin o ggeru sekai kakkoku no hitobito ga kono machi

1 13 6
Neimeehen ni atsumatte, aruku.
8
'More than 20,000 people from various countries throughout
the world gather in the town of Nijmegen and walk.'
Grammar = 4 points
vocabulary = 1 point

Kanji = 1 point
Length = 14 words

4. ggggggi mo nai.
'There are no winners or losers.’
Grammar = 0 points
Vecabulary = 1 point

Kanji = 2 points
Length = 3 words

5. ABHLLL mo nai.
'There is no order of placement.'
Grammar = 0 points
Vocabulary = 1 point

Kanji = 2 point
Length = 3 words

6. Yokkakan arukitooshita mono ni Kappa no gnggsho to
11 13 6
medaru ga ataerareru dake da.
9 15

'It's just that a certificate of completion and a medal
are given to those who last the distance over four days.‘

246

Grammar = 5 points
vecabulary = 3 points
Kanji = 2 points
Length = 13 words

Ketoshi no taikai ni wa nihon kara gojuurokunin sanka
shita.

“Fifty-six Japanese participated in this year's march.’
Grammar = 0 points

Vocabulary = 2 points

Kanji = 0 points
Length = 11 words

Sanka kyuukai no Kaneko Igmgggzg san (Nihon.Aruke.Aruke
‘fizggkaichoo) ni wa konna omoide ga aru.

3
'Tomokazu Kaneko, chairman of the Japan Walk-Walk Society,
who has participated nine times, has the following
recollection.
Grammar = 1 point
Vocabulary = 1 point

Kanji a 3 points
Length = 16 words

Hajimete sanka shita toki, nakama to issho ni "faito,
13
faito" to 5233135; ni 2252213 aruita.
12 8

'The first time he participated, he shouted with other
participating Japanese, "Fight! Fight!"'

Grammar = 3 points

Vocabulary = 2 points
Kanji = 3 points

247

Length = 15 words

10. Suru to, DOQEQL no josei ga suggmidete itta.
7 8
'Then an old woman came forward and said:'
Grammar = 2 points
Vocabulary = 2 points

Kanji = 2 points
Length = 8 words

11. "Naze‘gagggaou to in no."
4 12 2
"Why do you talk about fighting.”
Grammar = 4 points
Vocabulary = 0 points

Kanji = 1 point
Length = 5 words

12. "Watakushitachi wa [taggkai wa gomen] da wa."
”we don't want to fight."
Grammar = 0 points
Vecabulary = 2 point

Kanji = 1 point
Length = 7 words

13. Ganbarou to iikikaseru{gargi.gngggni_no "faito" datta

4 12 11 10 13 6

ga, Kaneko san we "we don't want to fight" to in kotoba ni

7 12 13

248

konorareta mono o sasshite, fukaku atama o sageta.

9 13 8
'It was a light "Fight!" calling on people to exert an
effort, but Kaneko realized the thinking behind the words,
"we don't want to fight," and hung his head in shame.’
Grammar = 12 points
Vocabulary = 2 points
Kanji = 4 points
Culture = 1 point
Length = 24 words

14. Aruku koto no daisuki na hitobito ga atsumatte, aruku
2 6 8
koto o tanoshimu.
2

'People who really like walking, gather and enjoy their
walking.‘

Grammar = 4 points

Vocabulary = 0 points

Kanji = 0 points
Length = 12 words

15. 11192;}; no aka 0 aruku.
1
'They walked over green hills.’
Grammar = 1 points
Vocabulary = 0 points

Kanji = 1 point
Length = 5 words

16. gpgazoi no michi o aruku.

1

249

'They walked on roads alongside canals.’
Grammar = 1 points
Vocabulary = 1 point

Kanji = 2 points
Length = 5 words

17. Arukinagara {areal ga umareru.
8 9
'While walking, contacts are made.’
Grammar = 2 points
Vocabulary = 1 point

Kanji = 0 points
Length = 4 words

18. "Kiiroi ribon, " ”Oosuzanna, " "Moshi, moshi 35.; yo" to
12
samazama na uta ga kikoete kuru.
.8

'Various songs, such as, "Tie a Yellow Ribbon,” "Oh,
Suzanna," and ”Moshi, moshi Kameyo", can be heard.‘

Grammar = 2 points
Vocabulary = 1 points

Kanji = 1 point
Length = 14 words

19. 929;; o utainagara neigai ga aruku.

8
'Soldiers walk while singing military songs.‘
Grammar = 1 point
Vocabulary = 1 point

Kanji = 4 points
Length = 6 words

250

20. Kuroi hada o gaggggkaseta musumesan ga aruku.
10 13
'Girls with glistening tanned skins walk.'
Grammar 8 2 points
Vocabulary = 1 point

Kanji a 1 point
Length = 7 words

21. Ryukku 0 [so ni shita] gggjin ga aruku.
13
'An old man with a knapsack on his back walks.‘
Grammar = 1 point
vecabulary = 3 points

Kanji = 2 points
Length = 8 words

22. Orinpikku no ggigfiig wa hang; yori mo kochira ni
16 1
uketsugarete iru no kamoshirenai.
11 9 2 14
'The real spirit of the Olympic Games may have been
preserved more in this (four days' march) than in the
Olympics itself.‘
Grammar = 6 points
Vocabulary = 3 points
Kanji = 3 points
Length = 13 words
23. ggdoo ni wa, isu o narabe, yasunde ike to tqlgggki o

8 8 4 12

251

suru shufu ga iru.

13

'Along the road, there are housewives who line up chairs
and beckon to walkers to come and rest.‘

Grammar = 5 points
vocabulary = 2 points

Kanji = 2 points
Length = 15 words

24. Koohii ya kukkii o furumau ie mo aru.
13

'There are families which treat (walkers) to coffee and
cookies.‘

Grammar = 1 points
Vocabulary = 1 point

Kanji = 0 points
Length = 8 words

25. Kuroobaa no hangggng o okutte kureru shoojo mo iru.
8 13

'There are girls who present (the walkers) with clover
bouquets.‘

Grammar a 2 points
Vocabulary = 1 point
Kanji = 1 point
Length = 9 words

26..Mainen no koto na no de, jun-in to sankasha wa Lgn_de
8
mbareru 1 to in .
9 12

'Since it is an annual event, (they) say that the

252

residents and participants are strongly tied together.‘
Grammar = 3 points
Vocabulary = 3 points

Kanji = 2 points
Length = 15 words

27. Ima nihon de wa, 'aruki' no gigggpha ga gggiggg shi,
8
yaku kyuuhyakuman to mo iwarete iru.
12 9

'In Japan today, those who like to walk have increased
drastically and are said to number about 9 million.‘

Grammar = 3 points
vocabulary = 2 points
Kanji = 4 points

Ten = 1 point

Length = 16 words

28. "walk" to in senmon zasshi mo dete iru.
12 13
'There is now a special magazine, "walk".'
Grammar = 2 points
Vocabulary = 0 points

Kanji = 3 points
Length = 7 words

29. Uhdoo buggfin o figigngg_suru tame.

13
'To provide a solution to the lack of exercise.’
Grammar = 1 point
Vocabulary = 3 points

Kanji = 3 points
Length = 6 words

253

30. Tairyoku to ggjgnjnryoku no ggnfigl ni igggg tame.
6 13
'To challenge one's physical and mental limits.’
Grammar = 2 points
vocabulary - 4 points

Kanji = 5 points
Length = 8 words

31. 592122 no 1522;; no tame.
6
'To nurture the spirit.‘
Grammar = 1 points
Vocabulary = 2 points

Kanji = 4 points
Length = 5 words

32..Arukitsuzukeru riyuu wa samazama da ga, aruki no 221212;
11 13 7
we yahari, aruku koto o tanoshimu koto daroo.
2 2 14

'There are various reasons for continuing to walk, but the
essence of walking is in its enjoyment.‘

Grammar = 6 points
vecabulary = 2 points
Kanji = 2 points
Length = 17 words
33. Tanoshiminagara arukeba, kaze no iro ga miete kuru.
8 8 8

'If you walk while enjoying it, you can see the color of
the wind.‘

Grammar = 3 points

2541

VOcabulary = 0 points
Kanji = 0 points
Ketsu = 1 point
Length = 8 words

255

1.

Hoogen to K yootsuugo

Dialects and the Common Language

Nihon wa, kuni ga semai war! n1, tochi ni yoru hoogen no
13 13
chigai ga ookii.

'For a small country, Japan has large differences in its
geographical dialects.‘

Grammar = 2 points
vocabulary = 2 points

Kanji = 0 points
Length = 15 words

Nihon no chiggi_wa yama ya kawa ga ooku,.lotoloto
8

kootsuu ni fuben de aru ue, mukashi wa, nihyaku suujuu mo

13 1
no claim no ryoochi ni wakarete 12111, ga okonawarete
6 9 8 9
ita.
'As for Japan's physical geography, in addition to there
being a lot of mountains and rivers naturally making it
difficult for transportation, in former times, Japan was
divided into 200 to 300 fiefs, owned and administered by
feudal lords.‘
Grammar = 7 points
Vocabulary = 6 points

Kanji = 4 points
Length = 31 words

256

3.

Sore de, hitobito wa, tagai ni jiyuu ni ikiki suru koto
7 2
ga muzukashikatta.
4

'So it was difficult for people to freely associate with
each other.‘

Grammar = 3 point
Vocabulary = 1 point

Kanji = 0 points
Length = 13 words

Sono tame, kotoba mo, hitotsuhitotsu no chihoogoto ni
7
wakareru yoo ni natta no de aru.
9 13 2

'For this reason, language, too, came to be divided into
individual geographical areas.’

Grammar = 4 points
Vocabulary = 1 point

Kanji = 0 points
Length = 15 words

Toku ni tooi chihoo to no ikiki wa taihen na koto datta
1 6

kara, hyakunen hodo mae made wa, Aomori no hito to

8 6

‘figggobi-a no hito ga deatte mo, hotondo kotoba ga

11 8
tsuujinakatta to in koto de aru.

4 12 2

257

8.

'Especially because association with a faraway district
was difficult, up to 100 years ago, even if a person from
Aomori and a person from Kagoshima met, it is said they
could not communicate.

Grammar = 9 points
VOcabulary = 3 points
Kanji = 2 points
Culture = 2 points
Length = 37 words

Konnichi de mo, sono tochitochi no hoogen wa, mainichi no
6
seikatsu no naka de ikiiki to tsukawarete iru.
1 9

'Even today, the dialect of each area is being used
actively in the middle of everyday life.’

Grammar = 3 points
Vocabulary = 1 point
Kanji = 0 points

Ten = 1 point
Length = 18 words

Endan yoku tsukau aisatsu no kotoba o mite miyoo.
l3 8

'Let's look at words that are ordinarily well-used for
greetings.’

Grammar = 2 points
Vocabulary = 0 points

Kanji = 0 points
Length = 9 words

"Konbanwa" to in yoru no aisatsu o, "Oban de gozaimasu"

12 13

258

9.

"Oban ni narimashita" aruiwa, tada "Oban" to in tochi ya

6 12 13

aru.

'There is a region where the evening greeting "Konban wa"
("Good evening.”) is said ”Oban de gozaimasu", "Oban ni
narimasu", or simply ”Oban”.

Grammar = 5 points

VOcabulary = 0 points

Kanji = 0 points
Length = 21 words

Mata, "Arigotoo" no kawari ni, "Ookini" to ka "Dandan" o
7 6
tsukau chihoo ga aru.
13

"Again, there is a district which uses "Ookini" and
”Dandan" instead of "Arigatoo" ("Thank you").

Grammar = 3 points
Vocabulary = 0 points
Kanji = 0 points
Length = 13 words

10. Korera no aisatsu kotoba ni wa, sono tochi sono tochi no

3 6 6
aji ga kalotte iru.

'As for these words of greeting, they are full of the
flavor of each particular region.‘

Grammar = 3 points
vecabulary = 1 point
Kanji = 0 points
Length = 15 words

259

11. Nazenara, hoogen wa, sono tochi no seikatsu ya rekisgi
7 6
to fukai tsunagari ga ari, kazokudoosbi,
1 8
to-odachidooshi, tochi no hitodooshi ga kutsuroida
6 13
hanashi o suru toki ni wa, shizen ni [kuchi o tsuite] dete
13 8 8
kite, tagai ni yoku kimochi o tsuujiau koto ga dekiru
8 2 13
mono dakara de aru.

'The reason for this is a dialect has a close relation
with the life and history of the local area, so when
family, friends, and people of that region engage in
relaxed conversation, the dialect just rushes out
naturally and is able to mutually convey emotion.‘

Grammar = 13 points
Vocabulary = 7 points
Kanji = 2 point
"Ketsu" = 1 point
Length = 47 words

260

Appendix B

The Passages Written in Japanese Orthography

Amke Orinpikku 'Walking Olympics'

 

 

U $U19>EV7

§¢.t7yroiéhmr.X$hfitfi
fibn5.711¥1x.ql%tbfidn.
$9”?ny 77.51%...hn1m6o
:EAtgiéfifiefioAvtficom+
4}l«yK%ior.$<.%Mmaofl
mttmowafifiéfiLkbomfifioflg
txrmfifiibnafinfi.
4$okamaa$mb£+fiAw§ML
kofimnfiofi¥§iakiafifififidm
éfivmu:bhfimwfibaoMbr§ML
kW.W@t.%Ka7741.7T4bot
fifimwkrfimk.
Tat.$RotfififiAwrmokoZa
€flsfiamfiooflkbuflwutbhfi
hefihaasamméweafima¥o
fi7741.._ fiokfi.‘ Amlicbfl <<o mos...
£2: 8 an? km5§ofimcbbhkbflé
§L1.fi<fitéfiko
$<ctokfl3hAUtflflior.$<

 

 

ct?§LD.fio&t$<olflfmofit
$3} fithflb$hbmxw$tnbo 258
myfiyafissxfy+a7991LCIat
ét¥itfifi8circéofifi2flmafl
bfififi$<.$hnfifflnakmahfi$
(.9a772fimLkéA#$<otvyE
w7®fiflfltfixbtcbbmfifiofinf
méobbLntm.

fifimu.4z&m«.W1rmnt$ms
tfaififim6.=leleav*ltia
t5i156.9axxromfitast<n
6wfi1m6.§$0cthor.EE&§m
fiuficfifinbtmfio

wraith. 2%.... oifififififik
fihfifiAt1mbhrw6. ..WALKa .c
m5§fififi1trmaofifififltflfifé
kbowfltfiflflomfimmUkbomfio
emokbofisficagmaétsttw.
fiémfifiutflfi.$<ct2§Lbctfi
65o .

etaamemna.moe#511<a.

 

 

26l

Hoogen to Kyootsuugo 'Dialects and the Common Language'

"l? '1“; 3‘) 1* 4.. o

.51 §wv MJMVE”%WU§wh
I 0 ' 1 A 0 p
w @142 HMfiTJsKw€=HrQJ<~:V“flQ
t$:V:”°w:P'<s:'”$3imaitmeku

M$Q$§A§°£°W€Q£'Mfio’[Olpggknw
ueésze”c°' Spew° wvHH; flkwgtmx
:”S(Qauw19”'4 m&flwfifivj ¢¢Q<w
' ganove'gwqhmm; QJfiQOg

112'1m31181v441
Vfir€300€mwwfivifi

i0)

rhonejmswewaew°”:4eeimommus'
wefiflwefifletaww°v3w°m$cw’ an' w
14111.1euw1; Oa1o1eo'11hme'wm
cwma'41149141v04sunm+4w422'm

’0

{E HUN”) )V-EVWV' QégflHVfifiJQJfillJfiuvhuw

O

A

221/#:1141414va

262

Appendix C

The Comprehension Questions
Amke Orinpikku 'Walking Olympics'

atomau.tޤac$mrtv.
o Efifitu.

2§a+bwfifiwflt5flkbflhfflfififib
fl‘Cufc.

O iilhtCJ: 675305 ”viii! L‘.

O

 

263

#3}: fdflACEECfi8¥T6 Z. tfltfl't‘bb‘o

"”1143.
53201839“. fificz‘flfc’a 7;.

:Efl-l—b 07(fi0) U J: ‘5 flfibfl'fl‘t’flfiéif
fibfl‘CL‘fc.

Bib. -’)—‘90fi75:t $250116 J: 5 K.
Rafa.

 

264

fitment—cu. fifimktfi W. 50A”
$3916.

HUfiEfflok.
fiwfifi«fi(®uflhkkhztfiak.

fltkggfififiUH09k.
HUEfiOAEok.

 

26S

{ointaoxgu
bottlflafuttb‘.
f0ifiifi0$§tfiu9t#0fimu.

’Efitfi E. 575“? Lfifia'cufxu
fiB‘DEEfl'fi‘CESESt fibfl'tub.

 

266

oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
....................................................................................
oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
oooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
.................................................

OOOOOOOOOOOOO
.............

OOOOOOOOOOOOO
.............

.............
OOOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOO
OOOOOOOOOOOOOOOOOOO
..................

OOOOOOOOOOOOO
OOOOOOOOOOOOOO

OOOOOOOOOOOOOO
OOOOOOOOOOOOO

OOOOOOOOOOOOO
.....

OOOOOOOOOOOOOOOOOOO
..................

:I;I;I:Z:I [ a 0 v t 5 j :‘I‘Z'PI'I‘I‘I‘I'I'I-Z‘I'I

.....
.....

.....
.....

[$5th 133323IEIEIEIEiEiEIEIEIEIEZ

.............
OOOOOOOOOOOOOO
.............

f 8 8 3 CC]

OOOOOOOOOOOOO
.............

l' ’5 A! ’5 All I;I;I;I;I;Z;I;Z;I;I:Z;I;I;I

OOOO

    

 

267

 

.....
.....

.....
.....

3U éflgitfl. til”? min 01:30)“!be o
‘CUI‘J’D.

fcfivCJ: ( antennas ltfi‘t‘fib.

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

35:17.41. am: utau‘r arc am».
1' 01:4? mflmltt Enact“ ‘) 5m».
inoAE 5 Lt‘ < abufc’fit but».

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

2:22:12: (flick previous {or the ptevious question.
::::::::::I (flick Next Passage to proceed to the next passage.

OOOO

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

268

  

ooooooooooooo
oooooooooooooo
ooooooooooooo
oooooooooooooo
ooooooooooooo
oooooooooooooo
ooooooooooooo
oooooooooooooo
ooooooooooooo
oooooooooooooo
ooooooooooooo
oooooooooooooo
ooooooooooooo
oooooooooooooo
ooooooooooooo
oooooooooooooo
.............
oooooooooooooo
.............
oooooooooooooo
.............
..............
.............

oooooooooooooo
.............
oooooooooooooo
ccccccccccccc
oooooooooooooo

ooooooooooooo
oooooooooooooo
.............
oooooooooooooo
ooooooooooooo
oooooooooooooo
ooooooooooooo
..............
.............
oooooooooooooo
ooooooooooooo
oooooooooooooo
ooooooooooooo
oooooooooooooo

OOOOOOOOOOOOO
OOOOOOOOOOOOOO
OOOOOOOOOOOOO
OOOOOOOOOOOOOO
.............

 

 
 

oooooooooooooo
ooooooooooooo
oooooooooooooo
ooooooooooooo

..............
OOOOOOOOOOOOOO
OOOOOOOOOOOOOO
OOOOOOOOOOOOOO
OOOOOOOOOOOOOO
OOOOOOOOOOOOOO
..............

Hoogen to K yootsuugo 'Dialects and the Common Language'

fihkfifiwfiit X filbb'Si 511310.
0 loch.
0 IBI’uflz‘fiEfibeA.

o gun taco/k.
o elm/vet.

 

269

 

3131333333 ezoxacu at?) Bflkfl‘im b t b 7.: b‘.

.....
.....

.....
.....

.....

5232313131 0 AA.

.....
.....

.....
.....

.....

5231325232 0 i+7tA .

.....
.....

.....
.....

.....

3233323231 0 1155A.

;:;:;:;:;: 0 :BA.

 

270

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

.............
OOOOOOOOOOOOOO
OOOOOOOOOOOOO

oooooooooooooo
.............
oooooooooooooo
ooooooooooooo

.............
OOOOOOOOOOOOOO
.............
OOOOOOOOOOOOOO
.............
OOOOOOOOOOOOOO
.............

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

OOOOOOOOOOOOOO

OOOOOOOOOOOOO
OOOOOOOOOOOOOO
OOOOOOOOOOOOO

 

 

5353315133 $E0tfififl é: mm: L f: b.

.....
.....

.....
.....

.....

3333333332 0 [774 |~. 774 H .

.....
.....

.....
.....

.....

11 Inflfihdbfituimj.

33$ raksuzo51u1.

OO

.....
.....

.....
.....

3133315353 rueaaa £5603] .

O

 

271

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

ooooooooooooo
oooooooooooooo
ooooooooooooo
oooooooooooooo
ooooooooooooo
oooooooooooooo
ooooooooooooo
oooooooooooooo
.............

.............
OOOOOOOOOOOOO
.............
OOOOOOOOOOOOO
OOOOOOOOOOOOOO
.............
.............
.............

 

#11::th bub—97:11».
o 305%!“ .
O zfifumiutt

O a3~fll>9k.
o $<ztti§blufcx

 

272

 

.....

flfi 9tflfl§fioffifltbkb.

flfi <370-n—o11.

5325232325 0 4 1 -

fig <3:—t—.

2:22:22: 0 a a 1 —.

 

2H

OOOOOOOOOOOOO
.............
.............
OOOOOOOOOOOOO
OOOOOOOOOOOOO

 

ismalufi‘ctv.
o Elfiitmfi‘tb z t.
o fibmllllcnczt.

o $<ztt§LCzL
0 1130351631.

 

 

274

BIBLIOGRAPHY

Alfonso, A. 1966. W. Sephia
University: Tokyo.

Ahmad, K. G. Corbett, M. Rogers, and R. Sussex. 1985.

 

Cambridge University Press.

Allen. James. 1987- W- Menlo

Park, CA: Benjamin Cummings.

Alty. J- L. and M. J. Coombs. 1984- W
agg_§;gmpleg. Manchester: NCC.

Anderson, R. C. and P. Freebody. 1981. Vocabulary knowledge.

. ‘ a 1;. Guthrie,
J. (ed. ), 77-117. Newark, DE: International Reading
Association.

Andersson. B- V- 1981 W
W. Unpublished doctoral

dissertation, University of New Orleans.

 
   
 

v

Andersson, B. V. and J. P. Gipe. 1983. Creativity as a
mediating variable in inferential reading comprehension.

Beading Psychology 4, 313-325.

Asahi Shinbun Ronsetsu Iinshitsu. 1985. Aruke orinpikku.

WWW. Tokym
Genshoboo.

Barnitz, J. G. 1986. Toward understanding the effects of
cross-cultural schemata and discourse structure on second

language reading comprehension. Journ§1_gfi_§§§ging_
fighgyig;,18, 95- 116.

Barlett, F. C. 1932. ° ' ' - ' .
§Qgi§l_g_ygn_;ggy_. Cambridge: Cambridge University Press.

Bloch, B. 1946a. Studies in colloquial Japanese I:

inflection. J9nrna1_9I_1he_Ameni£§n_Qriensal_§29121¥J
66, 97- 109.

 

----. 1946b. Studies in colloquial Japanese II: syntax.
nggugge 22, 200-248.

--—-. 1946c. Studies in colloquial Japanese III: derivation

27S

of inflected words. MW].
m. 66, 304—315.

--—-. 1950. Studies in colloquial Japanese IV: phonemics.
L5399§9§.25: 85-125-

Block, Ellen. 1986. The comprehension strategies of second

language readers. IE§QL_Qg§;te;1y 20, 463-494.

Bloomiield, Leonard. 1933. Language, New York: Holt,
Rinehart, and Winston.

Bork. Alfred. 1981- W. Bedford. MA:
Digital Press.

----. 1985- W- New York:

Harper & Row, Publishers, Inc.

----. 1987. The potential for interactive technology.
Byge_12.2, 201-206.

Bramble, W. J., E. J. Mason, and P. Berg. 1985. Computergdyn
School . New York: McGraw-Bill.

Brand, Steward. 1987. WI.

MIT, New York: Viking.

Brend. Ruth M. (ed.)- 1972. W
Wri ti ngg . The Hague: Mouton.

----. (ed.). 1974. agggnge_in_1§gm§mig§, in North-Holland
Linguistic Series. Amsterdam: North-Holland.

---- and Kenneth L. Pike (eds.). 1976. Iggmemigg. The Hague:
Mouton.

Buxton, Bill. 1990. Smoke and mirrors. By§e_15.7, 205-210.

Cameron, K. C., W. S. Dodd, and S. P. Q. Rahtz. (eds. ). 1986.

QQmD2L§I§_én__MQQ§IE_LQBQB§9§_§§nQi§fl. Chichester, WBBt
Sussex, England: Ellis Horwood Limited.

Canale, Michael and Graham Barker. 1986. How creative
language teachers are using microcomputers. IESQL,

HQ!§1§£E§I.20: 1-3-

Carrell, Patricia L. 1987. Introduction. In Beggargn_in
WWW- Devine. J - .
P. L. Carrell, and D. E. Eskey (eds.), 1- 7. washington,

276

D.C. 8 TESOL.

Carrell, Patricia L., Becky G. Pharis, and Joseph C. Liberto.
1989. Metacognitive strategy training for ESL Reading.

IEEQL.Q§§I§§I1¥.23: 547-573-

Carrell, Patricia L. and Joan C. Eisterhold. 1983. Schema

theory and ESL reading pedagogy. IE§QL_QQ§;§§;1y 17, 553-
573.

Carroll, J. B., P. Davies, and B. Richman. 1971. fig;g_
Egegggngy_nggh, Boston: Houghton Mifflin Company.

Chall, J. S. 1958. Eggggpi11gy, Columbus: Ohio State
University.

Chapelle, Carol. 1989. Using intelligent computer-assisted
language learning W 23 59- 70.

---- and Joan Jamieson. 1986. Computer-assisted language
learning as a predictor of success in acquiring English as

a second language. IE§QL_Qgg;;§;1y 20, 27-46.

Charniak, E. and D. McDermott. 1985.13;;g§gg;193_§g,
ALL-1W Reading. MA: Addison-Wesley-

Chaudron. Graig- 1988- §e99ng_Langnage_§la§§reem§1_3e§eargh.
q9n41§§gn1ngL§ng_L§§;n1ng, New York: Cambridge University

Press 0

Chomsky, Noam. 1957. fiyn;gg§1;_§§;ng§n;§§, The Hague: Mouton.

—---- 1965- Aa2e21g_91.1he_The9:1_91_§xntazy Cambridge. MA:
MIT press.

---- o 1982 o _ __ ._ ‘ -_‘
W- Cambridge. MA: MIT Press.

----. 1984. Le21urea_2n.99121n1391_ang_ainningy Dordrecht:

Foris Publications.

 

Coady, James. 1987. Comments on Strother and Ulijin. In

 

Devine, J., p. L. Carre11,and D. E. Eskey (eds. ), 101-
103. Washington, D. C. : TESOL.

Coleman, E. B. 1971. Developing a technology of written
instruction: some determiners of the complexity of prose.
. _ W

 

277

1n§1139119n, Rothkopf, E. z. and P. E. Johnson (eds.),
155-204. New York: Columbia University, Teachers College
Press.

Connor, Ulla and Robert B. Kaplan. 1987. _11§1ng_A§;Q§§,

WW Reading. MA: Addison-
Wesley Publishing Company.

Culley, Gerald R., G. G. Mulford, and J. Milbury-Steen. 1986.
A foreign language adventure game: progress report on an
application of AI to language instruction. QALI§Q_QQQ£QQL
4, 69—94.

Dever, Susan Young. 1986. Computer assisted reading

instruction (CARI)- In W

19;__g§ggm1g_zg;pggg§_, Dublin, Fraida, David E. Eskey, and
William.Grave (eds. ), 183- 216. Reading, MA: Addison-

wesley.

Devine, Joanne, Patricia L. Carrell, and David E. Eskey

(eds )- 1987. W
W- Washington, D. C.: TESOL.

Ebbitt, Wilma R. and David R. Ebbitt. 1982. £111§;;§_§g;gg.

gng_lnggz_;g_zngl1§n, Glenview, IL: Scott, Foresman, and
Company.

Eggington, William G. 1987. Written academic discourse in
Korean: implications for effective communication. In

't' ro s l ' LZ_T§31, Connor,
Ulla and Robert B. Kaplan (eds.). 153-168. Reading, MA:
Addison-wesley Publishing Company.

Ellis, Rod- 1986 W
Aggg1§111§1gn_. Oxford: Oxford University Press.

Ercoli, Paolo and Robert Lewis. 1988. 511111g1§1_1n13111g_gg§_
1991§_1n_figggg_1gn_. Amsterdam, The Netherlands: North-

Holland.

Eskey, David E. 1986. Theoretical foundations. In T§g_n1gg,

 

Fraida,DavidE. Eskey, and William Grave (eds. ), 3- 23.
Reading, MA: Addison-wesley.

Farmer, A. K. 1980a. Introduction. In W9;31ng_ggp§;§_1n,

n uist c T e a ssues n a an
L1ggg1§11g§_. Otsu, Y. and A. Farmer (eds. ), 3—27.

Department of Linguistics and Philosophy, MIT, Cambridge,

278

MA.

----. 1980b. Speculations on the interaction of morphology
and syntax: a look at complex verbs in Japanese. In
k' a i L' n u st eore ca 8 s

W Otsu. Y. and A- Farmer (ed8-).
29-39. Department of Linguistics and Philosophy, MIT,

Cambridge, MA.

----. 1980c. Reply to S. Kuno (1980). In WQ;king_2§2§I§_iB

WW
L1ggg1§§1_§, Otsu, Y. and A. Farmer (eds. ), 55- 61.

Department of Linguistics and Philosophy, MIT, Cambridge,
MA.

----. 1981. modularity- In Winn

f e J in isti s ote Pa ers Vo
Farmer, A. K. and C. Kitagawa (eds.). 9-24. Department of
Linguistics, University of Arizona, Tucson, Arizona.

----- 1984. W
Engl1gn, Cambridge, MA: MIT Press.

-—-- and C. Kitagawa. (eds. ). 1981a Egggggg1gg§_gfi_;g§,
1 _-O . er ‘1,- 0 J- 911: - .11 t .‘ o O = a9-
V01= 2. Department of Linguistics, University of Arizona,
Tucson, Arizona.

----. 1981b. Introduction. In W

ferenc 0 Ja an stics e e 3 Vol.
Farmer, A. K. and C. Kitagawa (eds. ), 3-7. Department of
Linguistics, University of Arizona, Tucson, Arizona.

Fillmore, Charles J. 1968. The case for case. 931gg1§11§_1n,
L1Qgg1§11g_mhgggy,, ed. by E. Bach and R. Harms, 1- 88. New

York: Holt, Rinehart and Winston.
Findley, Charles. 1986. Interactive videodisc: a powerful new

technology in language learning. TE§9L_Ngg§1g1§g; 20, 10-
11.

Fries, Charles C. 1963. L1ngg1§11g§_§ng_3§§g1ng, New York:
Holt, Rinehart and Winston.

Fisher, Scott S. and Jane Morrill Tazelaar. 1990. Living in a
virtual world. By;§_15.7, 215-221.

Freebody P. and R. C. Anderson. 1981.E11_91§__1_g111§;1gg,

0 one _'_fo; .go 91'. _o! o 0 7 -.-..._:§ '9

279

gomprehens1gn (Tech Report No. 162). Champaign: IL:
University of Illinois, Center for the Study of Reading.

(ERIC Document Reproduction Service No. ED 201 992)

Gagne. R. 1977- Wing. New York: Holt.

Rinehart and Winston.

Geisler, Cheryl, Alexander Friedlander, and Christine
Neuwirth, and David S. Kaufer. 1986. WARRANT: a
pedagogical environment for critical reading, reasoning,

and writing- W 20. 12.

Geoffrion, Leo D. and Olga P. Geoffrion. 1983. 99mpg;_;§_§ng_
Begg1ng_1g§§;gg§1_n, Reading, MA: Addison-wesley.

Gleason. 8- A- 1961 W
L1_gg1§11g_, New York: Holt, Rinehart and Winston.

Goodman, Kenneth S. 1967. Reading: a psycholinguistic

guessing game W 4 126-
135.

---- and Yetta M. Goodman. 1980. L1ngg1§11ggL

Agngggggg_a1b11gg;§pny, Newark, Delaware: International

Reading Association.

Grabe, William. 198L Contrastive rhetoric and text-type

research- In W L2.
1§1§_. Connor, Ulla and Robert B. Kaplan (eds. ). 115-137.

Reading, MA: Addison-Wésley Publishing Company.

---- and Robert B. Kaplan. 1989. Writing in a Second
Language: Contrastive Rhetoric. In B1gnn§§§_1n,fl;111391

Empgw§11ng_3§1_§;gg§g_§. Johnson, Donna M. and Duane H.
Roen (eds.), 263-283. New York: Longman.

Guthrie. J- T- (ed ) 1981. W
,R_§gg;gn_ggx1gg§,. Newark, DE: International Reading
Association.

Hainline, Douglas (ed. ). 1987. eW’ ve 0 nts i

W London: Cream Helm-

Hale, K. 1980. Remarks on Japanese phrase structure: comments

on the papers on Japanese syntax. In WOrk1ng Papggg 1n
' uist' o e ' sues in J anes

L1ngg1§§1g§, Otsu, Y. and A. Farmer (eds.), 185-203.
Department of Linguistics and Philosophy, MIT, Cambridge,

280

MA.

Hasegawa, N. 1981. Lexicalist grammar and Japanese passives.
In 2r99ee_inga_9f__he_Ariz2na.99nieren_e_9__flspane§e.

L1gggis§1cs, Coyote 2a pegg, V01 2. Farmer,.A. K. and C.
Kitagawa (eds.), 25-40. Department of Linguistics,
University of Arizona, Tucson, Arizona.

Hatch. Evelyn- 1982 MW
Appl1gg_L1ggg1§11g§_. Rowley, MA: Newbury House Publishers.

Heaton. J- B. 1985. W London:
Longman.

Higgins, John. 1983. Can computers teach? QAL199_ggg;n11 1.2.
4-60

----. 1987. Artificial Unintelligence: computer uses in

language learning. IB§QL_931;;§;12_21, 159-165.

---- and Tim JohnS- 1984- Wing.
Reading, MA: Addison-wesley.

Hinds, JOhn. 1983a. Contrastive rhetoric: Japanese and
English. Igg;_3, 183—195.

----. 1983b. Retention of information using a Japanese style
of presentation. Paper presented at the annual meeting of
TESOL , Toronto .

----. 1987. Reader versus writer responsibility: A New

Typology- In W
Text. Connor, Ulla and Robert B. Kaplan (eds. ). 141-152.
Reading, MA: Addison-wesley Publishing Company.

----. 1990. Inductive, deductive, quasi-inductive: expository
writing in Japanese, Korean, Chinese, and Thai. In

W
£e;_pg_§1xg,. Connor, Ulla and Ann M. Johns (eds.). 89-109.

Alexandra, VA: TESOL.

Imai, Takashi and Mamoru Saito (eds.). 1986. Sggg1g§_1n_
WWW
L1ggg1§11g§, Dordrecht, Holland: Foris.

Ishimori. Nobuo (ed-). 1977- www-

Tokyo: Koomura Tosho Shuppan Kabushikigaisha.

Jaeggle, O. A. 1981. A modular approach to ”passives". In

28]

   

:10 : 7 1 ;. ‘7: ‘-
L1ngu1st1gs , Coyote Papggs. V01. 2- Farmer, A. K. and C.
Kitagawa (eds. ), 41- 59. Tucson, Arizona: Department of
Linguistics, University of Arizona.

James, Mark O. 1987. ESL reading pedagogy: implications of
schema-theoretical research. In Begggggn_1n_3§§g1ng_1n

W- Devine, J., P- L- Carrell

and D. E. Eskey (eds. ), 175-188. Washington, D. C. : TESOL.

Johns, Christopher and Sue Fortescue. 1987. g§1ng_ggmpg§e;§,
W New York: Longman-

Johnson, P. 1981. Effects on reading comprehension of
language complexity and cultural background of a text.
15, 169-181.

--—-. 1982. Effects on reading comprehension of building
background knowledge. IESQL_Qg§;§§;1y_16, 503-516.

Joos. M- (ed.)- 1966. W 4th ed-

Chicago: University of Chicago Press.

Jung. Udo 0. H- 1988. W
Lgnggggg_1§§gn1ng, Frankfurt am.Main: Verlag Peter Lang.

Kenner, Roger. 1986. Approaches to creating courseware.
20, 9.

Kenning, M. J. and M-M. Kenning. 1983. An_1n1;gggg§1gn_19_
WW9. Oxford: Oxford

University Press.

Kintsch, W. and E. Green. 1978. The role of culture-specific
schemata in the comprehension and recall of stories.

DIEQQQI§§_£IQQ§§§§§.1: 1-13-
Kitagawa, C. 1981. Anaphora in Japanese: gage and 11213. In

P oceed'n izon Con e
L1gguist1cs, ngggg Eapggg. V01, 2. Farmer, A. K. and C.

Kitagawa (eds.), 61-75. Tucson, Arizona: Department of
Linguistics, University of Arizona.

Klare, G. R. 1974-1975. Assessing readability. 3&39139,
3W 10. 62-102.

Kosaka, M. 1980a. Subjacency in a grammar of Japanese. In
W 0 '-L: ‘ 1 J..!°L = _C: V . 1:0 = 8

W Otsu, I. and A- Farmer (edS- ).

282

115-125. Cambridge, MA: Department of Linguistics and
Philosophy, MIT.

--—-. 1980b. Reply to Kuno's comments. In flg;k1ng_zsps1s_
. 'n . . ue
L1sgs1s11ss, Otsu, Y. and A. Farmer (eds.), 133-136.
Cambridge, MA: Department of Linguistics and Philosophy,
MIT.

Rumor 5- 1973- Tne_5trnQ1nre_9f_1hs_lanane§s_1ansnas§a
Cambridge, MA: MIT Press.

----. 1980a. Comments on Lekach's paper. In fig;h1sg_zsps1s,
' L'n 'st' 8 V0 t'c s 'n Ja an
L1ngs1s11ss, Otsu, Y. and A. Farmer (eds.), 41-53.

Cambridge, MA: Department of Linguistics and Philosophy,
MIT.

----. 1980b. A note on Ostler' s non-transformational analysis

of Japanese case-marking. In fl_131sg_£sps1s_1n

u 'stic Issu 'n
L1sgs1ss1ss_. Otsu, Y. and A. Farmer (eds. ), 93-113.

Cambridge, MA: Department of Linguistics and Philosophy,
MIT.

----. 1980c. Comments on M. Kosaka's "Subjacency in a grammar
of Japanese". In flsgk1ng Papers 19 Lingsist1gs, 291, 2.

Theoreticsl Issgss in Japanese L1sgs1s§1ss. Otsu, Y. and
A. Farmer (eds.), 127-132. Cambridge, MA: Department of

Linguistics and Philosophy, MIT.

----. 1980d. A note on Tonoike' s intra-subjectivization

hypothesis - In W

reti sue J ane .Otsu, Y. and
A. Farmer (eds.), 149-156. Cambridge, MA: Department of
Linguistics and Philosophy, MIT.

----. 1980e..A further note on Tonoike' s intra-
subjectivization hypothesis. In fissg1sg_gsps1s_1n,

Ling2igr19s1_Y211_z1_Tne9ret19a1_1aaue§_in_ianenese.
L1sg_1s11gs, Otsu, Y. and A. Farmer (eds.), 171-184.

Cambridge, MA: Department of Linguistics and Philosophy,
MIT.

----- 1935- Anaphora in Japanese. Janansss.§xn1az_flerkah923
La Jolla, CA.

Kurahone, A. 1981. A categorial treatment of scrambling in
Japanese. In

283

WM Farmer. A- K.

and C. Kitagawa (eds.), 77-101. Tucson, Arizona:
Department of Linguistics, University of Arizona.

Kuroda, S.-Y. 1981. Some recent trends in syntactic theory

and the Japanese language. In Esoceegings of the Arizona

fere Ja a L' i t' s te ers V . .
Farmer, A. K. and C. Kitagawa (eds.), 103-121. Tucson,
Arizona: Department of Linguistics, University of Arizona.

Lado, Robert. 1972. Patterns of Difficulty in Vocabulary.

In Teash1a9_E2911§h_a§_a_§egond_1e999292» Allen, Harold 3-
and Russell N. Campbell (eds.), 275-288. Mew York: McGraw-

Hill International Book Company.

Last. Rex. 1984. W
Oxford: Basil Blackwell.

----- 1989- WWW
Laaga1ng, Chichester, west Sussex, England: Ellis Horwood

Limited.

Lawler, Robert W. and Masoud Yazdani. 1987. 511111g1a1

Iatsl11gesce and Eggcas1ss. Norwood, New Jersey: Ablex
Publishing.

Lee, Kai-Fu, Alexander G. Hauptmann, and Alexander I.
Rudnicky. 1990. The spoken word. By;a_15.7, 225-232.

Leech, Geoffrey and C. N. Candlin (eds.). gsmpg1a1s_1n_

W- New York:
Longman.

Lillie, David L., wallace E. Bannum, and Gary B. Stuck. 1989.
W Longman: New York-

Longacre. Robert E- 1964 W
E1s1g_Mansa1,. The Hague: Mouton and Company.

-—--. 1968- 9. '; . f *
WW 22114.11 Santa Ana.

Calif. : Summer Institute of Linguistics.

----. 1983. W. New York: Flemmi-

MacKay, Ronald, Bruce Barkman, and R. R. Jordan (eds. ). 1979.
W- Rowley. MA: Newbury House

Publishers.

 

2841

McAvinney, Paul. 1990. Telltale gestures. syss_15.7, 237—240.

McLaughlin. Barry- 1987 WWW
Lsa1n1ng,. Baltimore, MD: Edward Arnold.

Makino, Seiichi. 1991. Two teaching techniques: intensive
reading through identification of extended sentential
units and listening and writing practice through
dict-a-conversations. Paper Presented at The Third
Conference of the Lower Lake Erie Teachers of Japanese.
February 1-3, 1991. The University of Pittsburgh,
Pittsburgh, Pennsylvania.

Marantz, A. 1981a. Grammatical relations, lexical rules, and
Japanese syntax. In W
WW2 Farmer. A.

K. and C. Kitagawa (eds. ), 123-144. Tucson, Arizona:
Department of Linguistics, University of Arizona.

Marks, C. B., M. J. Doctorow, and M. C. Wittrock. 1974. WOrd
frequency and reading comprehension. Ina_1931na1_91_
W 67 259-263-

Martin, Gale, James Pittman, Kent Wittenburg, Richard Cohen,
and Tom Parish. 1990. Sign here, please. ny1a_15.7, 243-
251.

Martin, S- E- 1988- A_Eeferenge_§rammar_of_£anane§eg Rutland,
Vermont: Charles E. Tuttle Company.

Mauer. H- (ed.)- 1989- W
Heidelberg: Springer-Verlag.

Mikami, A. 1960. Zag_sa_flaaa_ga_flaga1, Tokyo: Kurosio
Syuppan.

Miller, R. A. 1967. Ins_1apansss_1angnaga, Chicago:

University of Chicago Press.

----- 1970- W. New Haven: Yale

University Press.

Mitchell, D. C. 1982. Ina_£199ass_gj_gsag1ag, New York: John

Wiley and Sons.

Miyagawa, S. 1980. Complex verbs and the lexicon. 99mg

Paps1s1_ygl_1, Department of Linguistics, Tucson,
Arizona: University of Arizona.

285

----. 1981. Paradigmatic structures and word formation. In

EIQ22EQifl9__QI_Eh2aEEHEHL.iLJflEflEHEELJHLJEflEflEflfii

L1ngu1st1cs, gsyo 1s Papsrs. V01. 2. Farmer,xA. K. and C.
Kitagawa (eds. ), 145-162. Tucson, Arizona: Department of

Linguistics, University of Arizona.

Mulford, George W. 1989. Semantic processing for
communicative exercises in foreign-language learning.

W 23. 31-44.

Nakajima, Kazuko. 1988. On developing Hypercard stacks for
study of Chinese characters: KanjiCard. QAL1QQ_QQn;naL
6' 75-870

N91900: A» N- 1974- Th__M9dern_Beader_§_1anane§ezzngli§n.

ghasagss1_21s11gna1y,, 2nd ed. Tokyo: Charles E. Tuttle
Company.

Newmeyer. Frederick J. 1980- W

New York: Academic Press.

-—--. 1988 W
W Cambridge:

Cambridge University Press.

Norman, D. A. and D. E. Rumelhart (eds. ). 1975.319191a11gss,
1a_ggga111ga,. San Francisco: W. B. Freeman and Company.

NYns, Roland R. 1990. An expert system in computer assisted

language learning- W 15 99- 103-

Oehrle, R. T. and Nishio, H. 1981. Adversity. In Egggaag1ngs
0 t1‘ ;_°.011 O1 e_-1 ‘ 0 :0-168‘ ,510Lf=ti 8 O O ‘

Paps1s1_ygl1_z, Farmer,.A. K. and C. Kitagawa (eds.),
163-193. Tucson, Arizona: Department of Linguistics,

University of Arizona.

Ostler, N. D. M. 1980. A non-transformational account of

Japanese case-marking and inflexion. In flggk1ag_2apa1s_1a
n ' t' V 8 Ja a e

L1sgs1s11ss, Otsu, Y. and A. Farmer (eds. ), 63- 91.
Cambridge, MA: Department of Linguistics and Philosophy,
MIT.

Otsu, Y. and A. Farmer. (eds.). 1980. WQ;k1ng_£apa;s_1a,
I' . t' M J 2 I] !' J I . I
L1sga1s11ss, Cambridge, MA: Department of Linguistics and
Philosophy, MIT.

286

Papert. Seymour. 1980. WWW
Eggs1131_1gsas, New York: Basic Books, Inc.

Partridge. D- 1986 W

W Chichester, West
Sussex, England: Ellis Borwood Limited.

Pearson. P- David (ed-)o 1984- W-
New York: Longman.

Pike, Kenneth L. 1943. Enssas1ss, Ann Arbor: University of
Michigan Press.

-—--. 1947a. Grammatical prerequisites to phonemic analysis.
3919 3, 155-172.

----. 1947b. £hgnam1ss, Ann Arbor: University of Michigan
Press.

----. 1948. Isna_1angsagss, Ann Arbor: University of Michigan
Press.

----. 1959. Language as particle wave and field. Ins_1sxas,
W 202037-540

----. 1967 W
W- The Hague : Mouton .

-—-- and Evelyn G. Pike. 1977. g;amma11sa1_gna1ys1s, Dallas:

Summer Institute of Linguistics.

----. 1983. W. Norwood, NJ: ABLEX Publishing
Corporation.

Poser. William J- (ed )- 1988 W
W Stanford:

Center for the Study of Language and information.

Pugh, A. K. and J. M. Ulijn (eds. ). 1984. Baag1ngs_figz,

O
o --- o ’ 00‘s‘ 0 -' o_ ’
--- g --- -- g . g -

and Foreign Language. London: Heinemann.

(C

Reinking, D. 1988. Computer-mediated text and comprehension
differences: the role of reading time, reader preference,

and estimation of learning. W111
23, 484-498.

Reinking, D. and R. Schreiner. 1985. The effects of computer-
mediated text on measures of reading comprehension and

287

reading behavior- WW 20. 536-552-

Reynolds, R. B., M. A. Taylor, M. S. Steffenson, L. L.
Shirey, and R. C. Anderson. 1981. W

W (Tech- Rep- NO- 201) Champaign. IL:
University of Illinois, Center for the Study of Reading.

Rich, Elaine. 1983. A111j1s1a1_1n1a111ganga, New York and

Toronto: McGraw-Hill.

Rosenbaum, Peter S. 1969. The computer as a learning
environment for foreign language instruction. £9;a1gn_

W 2: 457-455-

Rothkopf, E. z. and P. E. J0hnson (eds.). 1971. ysspal,
I n' E l I !l I l J E H '!!
1ns113911sn, New York: Columbia University, Teachers
College Press.

Rubin, Bela. 1986. Two approaches to CALL: curriculumpbased
courseware and language exploration activities. 13391,

HB!§1§EE§I.20: 13-14-

Rude. Robert '1'. 1986- Wra-
Englewood Cliffs, New Jersey: Prentice-Hall.

Rumelhart, David E. 1980. Schemata: the building blocks of
cognition. In Ih2QIQEiQQl_l_éEQ__in_B§§QiBQ_QQmQI§h_n§iQns
Spiro, Rand J., Bertram C. Bruce, and William E. Brewer
(eds. ), 33- 58. Hillsdale, New Jersey: Lawrence Erlbaum
Associates.

Rutherford, William.E. and Michael Sharwood Smith (eds. ).
1988. :1 . '
gsag1ngs_. New*York: Newbury House Publishers.

 

Saito, Mameru. 1985.

W PhD Dissertation. MIT.
Cambridge, Massachusetts.

 

Sakade, F., K. Emori, R. Friedrich, and S. Ohashi. 1959. ,a
W Tokyo= Charles E-

Tuttle Company.

Sapir, E. 1921. Languags, New York: Harcourt, Brace and
W0rld.

Schitai, Amdt. 1989. The design and development of an
interactive videodisc for foreign language learning.

288

W 29. 48-52-

Schreck, Richard. 1985. Artificial intelligence (AI) and
CALL: a new solution or an old problem?
20, 22-23.

Sells. Peter- 1985- W
Theories, Stanford: Center for the Study of Language and

Information.

Shen, Fen. 1989. The classroom and the wider culture:
identity as a key to learning English composition. College,

MW 40 4 459- 466-

Shibatani, M. 1976. Causativization. In gyg_az_§_g_§_man_ig§L
.9141 W Shibatani. M- (ed )

239- 294. New York: Academic Press.

Shibatani. M- (ed )- 1976 mm
W New York: Academic Press-

Singer. Harry and Dan Donlan- 1985- W
1g§_, Hillsdale, NJ: Lawrence Erlbaum.Associates.

Smith, Frank. 1971. gndg;g;ggding_3§§ging, lst ed. New York:
Holt, Rinehart and Winston.

—---- 1982- W. 3rd ed- New York: Holt:

Rinehart and Winston.

Spiro, Rand J., Bertram C. Bruce, and William E. Brewer

(eds ). 1980 W

Hillsdale, New Jersey: Lawrence Erlbaum.Associates.

Steffenson, M. 8., C. Joag-Dev, and R. C. Anderson. 1979.
A cross-cultural perspective on reading comprehension.

W 15. 10-29-

Steviok. Earl W- 1986. W
glggggggm, Cambridge University Press.

Strickland, Dorothy 8., Joan T. Feeley, and Shelley B.
Wepner- 1987 W

New York: Columbia University, Teachers College Press.

Strother, Judith B. and Jan M. Ulijn. 1987. Does syntactic
rewriting affect English for Science and Technology (EST)

text comprehension? In W
§§ggng_L§ngu§gg_. Devine, J., P. L. Carrell, and D. E.

289

Eskey (eds.), 89-101. washington, D.C.: TESOL.

Suppes, Patrick. 1966. The use of computers in education.

Sgientifig_amerigan.215 205' 220.

Thorndike. R. L. 1973 W
W. New York: Wiley-

Thurn, Martin. 1991. eDemonstration of the beginning course of
teract U d t J a e
Presented at The nThird Conference of the Lower Lake Erie
Teachers of Japanese. February 1-3, 1991. The University
of Pittsburgh, Pittsburgh, Pennsylvania.

Tonoike. S- 1979- MW
ereneee, Doctoral Dissertation, University of Hawaii.

----. 1980a. Intra-subjectivization. In W9;Ling_2epe;e
Lin is V . ssu
Linggieeiee,. Otsu, Y. and A. Farmer (eds. ), 137— 147.
Cambridge, MA: Department of Linguistics and Philosophy,
MIT.

-—--. 1980b. More on intra-subjectivization. In EQLKiBQ.

P rs in Lin t 8 I su 8
W Otsu, Y. and A- Farmer (eds ).

157-170. Cambridge, MA: Department of Linguistics and
Philosophy, MIT.

Ulijn, J. M. 1980. Foreign language research: recent trends

and future prosnectS- WW 3
17- 37.

----. 1984. Reading for professional purposes:
psycholinguistic evidence in a cross-linguistic
perspective. In ad n o ess onal Pur es:

WW Pugh A- K-8
and J. M. Ulijn (eds. ), 66- 81. London: Heinemann.

Underwood, John H. 1984. Ligggiegiee, Qomputers end the
Legggege_1eeene;, Rowley, MA: Newbury House.

----. 1989. On the edge: intelligent CALL in the 1990s.
QompnLera_and_tne_numaniriea.23 71- 34-

Vance. 'I'. 1987. MW-

Albany: State University of New York Press.

van Riemsdijk, Hank and Edwin Williams. 1986. lnLIQQQQLiQB_LQ.

290

Wm- Cambridge. MA: MIT Press-
Wardhaugh. Ronald. 1969- We.

New York: Harcourt, Brace and World, Inc.

Wegner, Etienne. 1987- WWW
Systems: Computational and gogpitive Approaches to the
WW- Los Altos, California: Morgan

Kaufmann Publishers.

Wilson, Kirk. 1986. ILIAD: an example of AI in language
instruction. T§§QL_flege1ep§ep. 20, 22—23.

Winograd Terry- 1983 W
_elpme_1;_§yp§e§,. Reading, MA: Addison-wesley.

----. 1984. Computer Software for working with language.

W 251, 130-145-
Winston, Patrick Henry. 1977. A;pi£ieiel_lppelligepee,

Reading, MA: Addison-wesley.

291

HICHIan STQTE UNIV. LIBRARIES
tlltlfllllllfllll!HIHIIHIIIIIHHIIIIMNHHIIHIWHIHI
31293008964615

 

* M'- ----—-—-—-—q