LEARNING SUPPORT FOR ENGLISH COMPOSITION
BY ASKING BACK QUESTIONS
Hiroki Hidaka, Yasuhiko Watanabe and Yoshihiro Okada
Department of Media Informatics, Ryukoku University, Otsu, Shiga, Japan
Keywords:
Learning support for English composition, Asking back question, Realizable possibility, Suppositive expres-
sions.
Abstract:
There are several gaps between Japanese and English expressions, such as suppositive expressions. These gaps
make it difficult for Japanese students to study English composition. For example, realizable possibilities are
described clearly in English suppositive expressions, on the other hand, they are frequently omitted in Japanese
suppositive expressions. As a result, when Japanese students translate Japanese suppositive expressions into
English, they are often forced to reveal the realizable possibilities which are not described clearly in Japanese
expressions. In this way, it is important to make students aware of realizable possibilities when they try to
translate Japanese suppositive expressions into English. To solve this problem, in this paper, we propose a
learning support method for English composition by using asking back questions. Our system asks users back
and makes them aware of realizable possibility.
1 INTRODUCTION
It is difficult for Japanese students to study English
composition because there are several gaps between
Japanese and English expressions. Take realizable
possibility in suppositive expressions for example. In
English sentences, realizable possibilities are clearly
expressed in suppositive expressions.
(ex 1) I’ll call you when I get to Narita Airport.
(ex 2) If I get to Narita Airport, I’ll call you.
(ex 1) shows that the speaker is sure to get to Narita
Airport. On the other hand, (ex 2) shows that the
speaker has a fifty-fifty chance of getting there. In
contrast, in Japanese sentences, realizable possibili-
ties are frequently omitted or expressed ambiguously.
(ex 3) Narita kuko (airport) ni (to) tsui (get) tara
(when/if) denwa (call) shimasu (will).
In this sentence, the possibility of getting to Narita
Airport is not expressed clearly. Both a man who is
sure to get to Narita Airport and a man who has a
fifty-fifty chance of getting there can speak (ex 3).
As a result, when Japanese students translate
Japanese suppositive expressions into English, they
are often forced to reveal realizable possibility be-
cause they are not described clearly in Japanese sup-
positive expressions (Figure 1).
Figure 1: The translation process of Japanese suppositive
expressions into English: 1. the estimation of realizable
possibility, 2. translation.
A considerable number of studies have been made
on English composition support by extracting En-
glish expressions from Web documents (Oshika 05)
(Takeda 94) (Yamamoto 99) (EDP 07). In these stud-
ies, however, little attention has been given to the gaps
between Japanese and English expressions. Suppose
that a Japanese student wants to translate (ex 3), how-
ever, does not know that the realizable possibility is
the key to translating Japanese suppositive expres-
sions into English. If (ex 1) and (ex 2) are given as
the translation examples of (ex 3) to the student, it is
difficult for the student to determine which sentence
is proper without the viewpoint of realizable possibil-
367
Hidaka H., Watanabe Y. and Okada Y. (2009).
LEARNING SUPPORT FOR ENGLISH COMPOSITION BY ASKING BACK QUESTIONS.
In Proceedings of the First International Conference on Computer Supported Education, pages 367-370
DOI: 10.5220/0001845103670370
Copyright
c
SciTePress
Figure 2: System overview.
ity. As a result, it is important to make students aware
of the gaps, in this case, the realizable possibility.
To solve this problem, we propose a learning sup-
port method for English composition by asking back
questions. Our system asks users back and make them
aware of the gaps between Japanese and English ex-
pressions. There are several kinds of gaps between
Japanese and English expressions. However, in this
paper, we have concentrated on suppositive expres-
sions because space is limited.
2 ASKING BACK QUESTIONS
ABOUT POSSIBILITY
From the viewpoint of realizable possibility, English
suppositive expressions can be classified into four
types:
Type A expressions about general or habitual activi-
ties and the possibility is very strong
(ex 4) When you mix red and yellow, you get or-
ange.
(ex 5) You always play baseball whenever the
weather is nice.
Type B expressions about one-time activities and the
possibility is very strong
(ex 6) You will play baseball when the weather is
nice.
Type C expressions about one-time activities and the
possibility is fifty-fifty
(ex 7) If the weather is nice, you will play base-
ball.
Type D expressions about one-time activities and the
possibility is very weak
(ex 8) If the weather was nice, you would play
baseball.
Because, in Japanese suppositive expressions, realiz-
able possibilities are frequently omitted or expressed
ambiguously, it is important to make Japanese stu-
dents aware of the realizable possibilities.
To solve this problem, our system asks users back
and make them aware of the gaps between Japanese
and English expressions. Figure 2 shows the overview
of our system. Our system applies morphologic anal-
CSEDU 2009 - International Conference on Computer Supported Education
368
ysis(Kurohashi 05) to an input sentence, obtains con-
ditional clause (basic and original form) and conse-
quence clause, generates four types of asking back
questions according to the rules in Figure 3 and gives
them to the user. Take (ex 9) for example.
(ex 9) moshi (when/if) tenki (weather) ga haretara
(nice), yakyu (baseball) ga dekiru (will play).
From (ex 9), our system extracts moshi (when/if)
tenki (weather) ga haretara (nice), as the condi-
tional clause, on the other hand, “yakyu (baseball) ga
dekiru (will play)” as the conclusion clause. Then, ac-
cording to the rules in Figure 3, our system generates
four types of asking back questions:
Asking back question (type A) [for general or ha-
bitual activities]
(ex 10) tenki ga hareru toki ha, itsumo yakyu ga
dekiru, desu ka? (You think it always happens
that you play baseball whenever the weather is
nice, don’t you?)
Asking back question (type B) [for very strong
possibility]
(ex 11) tenki ga hareru koto ha kakujitsu ni okoru
node, moshi tenki ga harereba yakyu ga dekiru,
desu ka? (You think it is certainly that the
weather will be nice and it certainly happens that
you will play baseball, don’t you?)
Asking back question (type C) [for fifty-fifty possi-
bility]
(ex 12) tenki ga hareru ka douka wakaranai ga,
moshi tenki ga harereba yakyu ga dekiru, desu
ka? (You think it is fifty-fifty that the weather
will be nice and are tentatively planning that you
will play baseball, don’t you?)
Asking back question (type D) [for very weak pos-
sibility]
(ex 13) tenki ga hareru koto ha arie nai ga,
moshi tenki ga harereba yakyu ga dekiru, desu
ka? (You think it is almost impossible that the
weather will be nice, however, you are dreaming
that you would play baseball, don’t you?)
Then, the user answers the asking back questions,
finds a gap between Japanese and English expres-
sions, and translates the Japanese expression into En-
glish by using explanations and example sentences
which are generated by our system and consistent
with the selected possibility. For example, Figure
4 (a) shows the explanation and example sentences
which our system gives to the user when he/she
chooses very strong realizable possibility. On the
(rule for type A) [for general or habitual activities]
[conditional clause (basic form)] toki (whenever) ha it-
sumo (always) [consequence clause] desu ka ? (You think
it always happens that [consequence clause] whenever
[condition clause], don’t you?)
(rule for type B) [for very strong possibility]
[conditional clause (basic form)] koto ha kakujitsu ni
(certainly) okoru (happen) node [conditional clause (orig-
inal form)] [consequence clause] desu ka ? (You think it is
certainly that [conditional clause] and it certainly happens
that [consequence clause], don’t you?)
(rule for type C) [for fifty-fifty possibility]
[conditional clause (basic form)] ka douka wakaranai
ga (fifty-fifty) [conditional clause (original form)] [con-
sequence clause] desu ka ? (You think it is fifty-fifty that
[conditional clause] and are tentatively planning that [con-
sequence clause], don’t you?)
(rule for type D) [for very weak possibility]
[conditional clause (basic form)] koto ha arie nai ga,
(impossible) [conditional clause (original form)] [conse-
quence clause] desu ka ? (You think it is almost impos-
sible that [conditional clause], however, you are dreaming
that [consequence clause], don’t you? )
Figure 3: Generation rules of Asking back question about
realizable possibility.
Explanation You want to compose English sup-
positive expressions with very strong realizable
possibilities. In such a case, you should not use
if clause.
Japanese Narita kuko ni tsui tara denwa shimasu
English I’ll call you when I get to Narita Airport.
(a) An explanation and example sentences for English
suppositive expressions with very strong realizable
possibilities.
Explanation You want to compose English sup-
positive expressions with fifty-fifty realizable
possibilities. In such a case, you should use if
clause.
Japanese Narita kuko ni tsui tara denwa shimasu
English If I get to Narita Airport, I’ll call you.
(b) An explanation and example sentences for English
suppositive expressions with fifty-fifty realizable
possibilities.
Figure 4: Explanations and example sentences which are
consistent with user’s selected realizable possibility.
other hand, Figure 4 (b) shows the explanation and
example sentences which our system gives to the user
when he/she chooses fifty-fifty realizable possibility.
LEARNING SUPPORT FOR ENGLISH COMPOSITION BY ASKING BACK QUESTIONS
369
3 EXPERIMENTAL RESULTS
We examined whether nine Japanese students under-
stood realizable possibility which was consistent with
the given situation. In this experiments, we gave the
following Japanese suppositive sentences
(input 1) 962 do (degrees centigrade) made (to)
kanetsushi (heat) tara (when/if), gin (silver) ha
tokeru (melt)
(input 2) shigoto (job) ga owat (over) tara (when/if),
renraku shimasu (get in touch)
(input 3) ano ki (the tree) wo kiritaoshi (cut down)
tara (when/if), motto (more) nagame (view) ga yoku
naru darou (be good).
and some situations of each input sentence (Figure 5)
to the students. Then, our system gave asking back
questions to the students and we examined whether
the students understood realizable possibility which
was consistent with the given situations. Table 1
shows the experimental results. In Table 1, underlined
numbers show the numbers of students who select
asking back questions which, we first thought, were
consistent with the given situations. As shown in Ta-
ble 1, students’ answers were almost the same as our
answers, except in situation 1-2 and 3-3.
In situation 1-2, we first thought that type C ask-
ing back question was consistent with situation 1-2.
However, four students selected type D asking back
question because they thought that their friends were
not specialists and it is impossible to heat silver above
900 degrees Celsius. On the other hand, in situation 3-
3, five students selected type D asking back question
which, we first thought, were consistent with situation
3-3. The reason why these five students thought the
possibility was very weak was that they thought they
could not cut down the tree in someone else’s garden.
In contrast, three students selected type C asking back
question. The reason why these three students thought
the possibility was fifty-fifty was that they thought the
tree would fall down naturally or somebody would cut
it down. In both cases, students’ answers were di-
vided and some students found the possibility which
was not consistent with what we expected, however,
consistent with what they thought. It shows the effec-
tiveness of our method.
situation 1-1 You are a science teacher. You will
tell the nature of silver to your students.
situation 1-2 You will give some advices to your friend
who intends to performing experiments.
(a) situations for input 1
situation 2-1 There are prospects of finishing your task.
situation 2-2 There are little prospects of finishing
your task.
situation 2-3 There are no prospects of finishing your task.
(b) situations for input 2
situation 3-1 You are rebuilding your house and have
already decided to cut down the tree.
situation 3-2 You are rebuilding your house and now
discussing whether you cut down the tree.
situation 3-3 You are taking a walk and watch the tree
in someone else’s garden.
(c) situations for input 3
Figure 5: Situations for input 1, 2, and 3.
Table 1: Experimental Results.
asking back situation
question 1-1 1-2 2-1 2-2 2-3 3-1 3-2 3-3
type A 6 0 1 0 0 0 1 1
type B 2 0 8 1 0 9 0 0
type C 1 5 0 8 2 0 8 3
type D 0 4 0 0 7 0 0 5
REFERENCES
Oshika, Satou, Ando, and Yamana: An English Com-
position Support System using Google, IEICE
DEWS2005, 2005 (in Japanese).
Takeda and Furugori: A Sample-Based System for Help-
ing Japanese Write English Sentences, Trans. of IPSJ,
Vol.35, No.1, 1994 (in Japanese).
Yamamoto and Kitamura: Courpus based natural language
processing and an education system using it, Trans. of
JSISE, Vol.16 No.1, 1999 (in Japanese).
Electronic Dictionary Project: EIJIRO 3rd Edition, ALC,
2007 (in Japanese).
Kurohashi and Kawahara: JUMAN Manual version 5.1,
Kyoto University, 2005 (in Japanese).
CSEDU 2009 - International Conference on Computer Supported Education
370