本文(Article)

福岡大学理学集報 39 ⑴ 35∼43(2009)
− 35 −
Dependence on Context
in case of English-Japanese Machine Translation I−9
Katsuyuki SHIBATA1)
( Received November 30, 2008)
英和機械翻訳における文脈依存性Ⅰ− 9
柴田 勝征1)
(平成20年11月30日受理)
Abstract
This is the nineth of a series of articles on the context dependency analysis in case of EnglishJapanese machine translation system which we call US system . In this issue we examine the examples taken from Lesson 9 of four English textbooks for the first year grade Japanese junior high
school students.
The main subjects to be discussed in this issue are the followings: (1) Should that , those be
translated into are , arera or sore , sorera ? (2) How to translate questions and answers starting
from a question using interrogative pronoun which , (3) Which is the better choice for interjection
well between eeto and sore-de-wa ? (4) Which is the better choice for Oh, yes! between Ee,
mochiron-desu! and Aa, sou-desu! ? (5) Selection among karera , kanojo-tachi , hito-bito and
sorera for they and them (revisited), (6) Replace the translation of the substitutive verb do by
that of the verb in the preceding sentence (in case of status verbs), (7) A second example where you
is interpreted as plural, (8) Selection of Japanese for then among sou sure-ba , sono toki , sorekara ,
sore-de-wa , suruto and sono koro ni-wa in case of inquiry sentence following a negative statement, and (9) Add the Japanese translation of the main verb of the preceding sentence to that of a
prepositional phrase.
Key words: machine translation, context dependence, context inherited from the preceding sentences,
US system.
1 .Should
that , those be translated into
are , arera or sore , sorera ?
In the first article
[1]
of this series, we
investigated criteria for deciding the Japanese
1)
translation of that between are and sore .
There we explained that in a comparative
context as This is ..... That is .... , we should
translate that as are . In contrast, In the fourth
article [2] of this series, we showed an examle in
Department of Applied Mathematics, Faculty of Science, Fukuoka University, 8-19-1, Nanakuma, Jonan-ku,
Fukuoka, 814-0180, Japan
福岡大学理学部応用数学科,〒814-0180 福岡市城南区七隈8-19-1
− 36 −
which that should be translated as sore . In
addition, in the fifth article [3], we treated the
case where that precedes a noun and plays the
role of a demonstrative article . In that case,
we explained that if the noun succeeding that
does not correspond to any nouns mentioned
beforehand, such that should be translated as
ano insted of sono .
This time we treat that which is situated at
the end of a question sentence with main verb
be .
See the following example sentences.
Tom: Whose computer is that ? ............ (1)
22;*あれ;2;LT;T0=?;T-1=v;T-1<>p;D-2=つ個本枚冊台
匹頭脚軒;DJ-2<>OJ$;DJ-2<>SJ$; K845;
The same way of reasoning holds for the
plural form of that , ie for those . But, as
mentioned above, that and those play not
only as demonstrative pronouns but also as
demonstrative articles when they proceed a
noun. Actually, for those in example (2) above,
the first choice rule applied is the one below
rejecting sorera no if those in (2) is regarded
as a demonstrative article (the corresponding part
of speech is then 8 insted of 2 which means a
pronoun).
トム「あれ は誰のコンピューターですか?」
Tomu Are wa dare no konpyuutaa desu ka ?
Kenji: It's my brother's.
健二「それ(そのコンピューター)は私の兄弟の
です.
」
Kenji Sore( sono konpyuutaa )wa watashi
no kyoudai no desu.
Tom: Whose computer games are those?
........................ (2)
トム「あれら は誰のコンピューターのゲームです
か?」
Tomu " Arera wa dare no konpyuutaa no
geemu desu ka ?"
The following choice rule confirms that that'
in (1) above is at the end of an interrogative
sentence, that that ' in question is proceeded
by be' verb (T-1=v;) which is not in the past
tense (T-1<>p;), that that be' is preceeded by a
countable noun whose corresponding counting
unit is either tsu', ko', hon', mai', satsu',
dai', hiki', tou', kyaku', or ken'. The rule
further checks that the noun preceeding be' in
( 1 ) has not appeared as the subject (SJ$) nor
the object (OJ$) of the preceeding sentence,
and thus concludes that that' in (1) should be
translated as are'. That' in this case does not
imply something mentioned earlier but points out
something seen from both the speaker and the
interlocutor.
88;それらの;2;E0=those;E1<>people;DJ1<>SJ$;DJ1<
>OJ$;DJ1<>SJ2$;DJ1<>OJ2$;E1<>*BS;E1<>*BS2;
D1<>回; v580;
But the subsequent analysis verifies that those'
in (2) is in fact a demonstrative pronoun, and the
same rule K845 above applies also to those' in (2),
and the Japanes translation for those' in (2) is
determined as arera'. We remark that in our US
system, the choice indication ;*are; demands to
select the Japanese candidates with are at the
head or at the taile as well as are itself.
Contrary to the cases explained above, if those'
is considerd to point out a noun mentiond in
the preceeding sentence, it will be translated as
sorera' or sorera no' insted of arera' or arera
no'. See the following example.
Mike: Is this a Japanese stamp, Ken ?
マイク「これは日本の切手ですか,健?」
Maiku
Ken ?
Kore wa Nihon no kitte desu ka,
Ken: Yes, it is. These are all Japanese stamps.
健「はい,そうです.これらはみんな日本の切手
です。
」
Ken Hai, sou-desu. Korera wa minna Nihon
no kitte desu.
Dependence on Context in case of English-Japanese
Machine Translation I−9 (K. Shibata)
Nancy: Are those Japanese stamps, too ?
.............. (3)
ナンシー「あれら も日本の切手ですか?」
Nanshii Arera mo Nihon no kitte desu ka ?
Ken: No, they aren ' t. They are Chinese
stamps. ...................... (4)
健「いいえ,ちがいます.それら は中国の切手で
す.」
Ken Iie, chigai-masu. Sorera wa Chuugoku
no kitte desu
As regards to 'those' in example (3) above, at
first the following choice rule
88;* そ れ ら の ;2;E0=those;J1=5;J2=1;T2=xX;E2=*BS;
v570;
is applied. This choice rule confirms that the
word succeeding those in question is an
adjective, (the part of speech code is 5 ) and
the adjective is further followed by a plural noun
(J2=1;T2=xX;), and that the same noun appears
in the preceeding sentence (E2=*BS;). Verifying
all these conditions, the rule selects sorera no
rejecting arera no . But this rule compares only
the demonstrative articles among the Japanese
translations of those . In the sequal, those in (3)
is determined to be a demonstrative pronoun and
the following rule
22;*あれ;2;T0=?;T-1=v;T-1<>p;T2=y; K840;
− 37 −
correspond to are , but is sometimes used in
the sence of sore . So the distinction between
these two usages of that gives rise to a very
difficult context-solving problem when it should
be translated into Japanese. By the way, the
interrogative pronoun which begins to appear
in Lesson 9 in high school English text books.
So we begin to examine the translation of which
in the next section.
2 . How to translate questions and answers
starting from an interrogation using interrogative pronoun which (part one).
Interrogative pronoun which should be
translated into Japanese as either dochira (when
two things are compared) or dore (when three
or more things are compared). In the example
below, there is a hint at the end of the sentence,
namely, this one or that one . Therefore we
understand that two things are compared and
decide that which in question should be
translated into Japanese as dochira .
Tom: Which is your game, this one or that
one?
トム「あなたのゲームはどちら ですか? これで
すか,それともあれですか?」
Tomu Anata no geemu wa dochira desu ka ?
Kore desu ka, sore-to-mo are desu ka ?
Kenji: This one is mine. .......................... (5)
健二「これが 私のものです.
」
finally determines the Japanese translation of
those in (3) as arera rejecting the competing
candidate sorera . This conclusion is correct.
But the reasoning is insufficient. As a matter
of fact, in the same way as is the case for that
mentiond in [1], those in (3) is in a comparative
context as These are .... Are those ....., too? ,
which implies that this those should be
translated as arera .
In Japanese as well as Korean, demonstrative
pronouns are classified in four categories as ko(kore, kono), so- (sore, sono), a- (are, ano), do(dore, dono). These correspond approximatively
to English pronouns this , it , that , and which .
But in the pragmatics, that does not allways
Kenji Kore ga watashi no mono desu.
Tom: What's its name ?
トム「それの(そのゲームの)名前は何といいま
すか?」
Tomu Sore no( sono geemu no )namae wa
nan-to ii-masu ka ?
Kenji: Well, its name is in the puzzle. Please
find the answer. ...........(6)
健二「ええと ,それの(そのゲームの)名前はパ
ズルの中にあります.答えを見付けてください.」
Kenji Eeto, sore no( sono geemu no )namae
wa pazuru no naka ni ari-masu. Kotae wo mitsukete kudasai.
− 38 −
The subject noun in a Japanese sentence
is followed by either postposition wa ' or
postposition ga'. It is usually very delicate and
difficult to choose between wa' and ga'. In
order to master the usage of wa' and ga', one
should learn various distinct cases and contexts
by heart. The usage of ga' in case of an answer
sentence to a question starting Which' is one
of them. For example sentence (5) above, the
following generation rule which is applied to its
translation
9;23;X;0;EB=Which ;J-1<>0?!;J-2<>0?!;T0<>h?;DJ2<>
「 ;DJ 3 <> 「 ;DJ 4 <> 「 ;E 2 <>and;T 1 <>w;J 0 #+ が :
ga;J1#VR;J1#DES; S935;
verifies that the preceeding sentence bigins
with Which ... , and thus adds the adequate
postposition ga to the subject pronoun (J0#+が:
ga;).
3 .Which is the better choice for interjection
well between eeto and sore-de-wa ?
Interposition well appearing at the head of
a conversational sentence has two Japanese
translations; namely the one, eeto , which is
to take pause before answering the question,
and the other, sore de-wa , which is to change
the topic of the conversation. In example (6)
in the previous section, well is at the head of
a sentence and the Japanese translation of the
preceeding sentence terminates with interrogative
postposition ka ? . Verifying these conditions,
the following choice rule decides that well in (6)
should be translated as eeto .
XX;*ええと;2;E0=Well;FT;E1=,;PJS=*か?」; x670;
4 .Which is the better choice for Oh, yes! be-
tween Ee, mochiron-desu! and Aa, soudesu! ?
Interjective sentence Oh, yes! has two usages;
namely, Ee, mochiron desu! , which is used as a
strongly affirmative answer to a question posed
by the interlocutor, and Aa, sou desu! , which
expresses speakers approval to the interlocutor s
assertion.
Meiling: It is a duck.
美玲「それはあひるです.
」
Meirin Sore wa ahiru desu.
Ken: It is a rabbit too.
健「それ(そのあひる)はウサギでもあります.」
Ken Sore( sono ahiru )wa usagi de-mo ari-
masu.
Meiling: Oh, yes! ................. (7)
美玲「ああ,そうです! 」
Meirin Aa, sou-desu !
In example (7) above, the following choice rule
XX;*ああ,そうです;2;E0=Oh , yes;BS=* too.; x545;
is applied. The rule checkes that the preceeding
sentence ends with ... too. and consequently
that the preceeding sentence is not an
interrogative sentence, and indicates Oh, yes!
as speaker s approval.
5 .Selection among
karera , kanojo-tachi ,
hito-bito and sorera for they and them
(revisited).
In the fifth article [3] of this series, we explained
that plural pronoun they and its objective
case them have Japanese translation karera ,
kanojo-tachi , hito-bito and sorera . We
presented there a choice rule which rejects
kanojo-tachi from among the Japanese
candidates if the sentence in the process of
translation contains no word with attribute W
(symbolizing woman ) either before nor after
they or them , and, further, if the preceeding
sentence nor the sentence before the last containe
no word with attribute W . In Lesson 9, the
same choice rule
22;彼女;2;T-1<>r;TA<>W;TZ<>W;SO<>W; E030;
is again applied to example (4) of section 1 at the
beginning of this article.
Dependence on Context in case of English-Japanese
Machine Translation I−9 (K. Shibata)
The context information (the information about
the preceeding sentence and the sentence before
the last) is also referred in such a context as
shown below.
Nancy: Do you know this girl, Ken?
ナンシー「あなたはこの少女を知っていますか,
健?」
Nanshii Anata wa kono shoujo wo shitte-imasu ka, Ken ?
Ken: Yes, I do. She is Kumi ' s cousin. I
know her very well.
健「はい,知っています.彼女は久美のいとこで
す。私は彼女をとてもよく知っています.
」
Ken Hai, shitte-i-masu. Kanojo wa Kumi no
itoko desu. Watashi wa kanojo wo totemo
yoku shitte-i-masu.
Nancy: Do you know these boys, too?
ナンシー「あなたはこれらの少年も知っています
か?」
Nanshii Anata wa korera no shounen mo
shitte-i-masu ka?
Ken: Yes, I do. They are Masao and Osamu.
............ (8)
健「はい,知っています.彼等は正夫(正雄,正
男)と治(修)です.
」
Ken Hai, shitte-i-masu. Karera wa Masao to
Osamu desu.
As regards to They' in example (8) above, the
following choice rule
22; 人 び と ;2;E0=they;T0=O;OZ=FxX;OZ=h;SZ<>Fx
X; J800;
is adopted to reject hitobito from among the
Japanese candidates. This rule verifies that
they in question begins with an uppercase letter
(T0=O;) which means this they is situated at
the head of the sentence. When the attribute set
of the subject (SZ) or of the object (OZ) of the
preceeding sentence contains human attribute
symbol h as well as plural attribute symbols
xXF , They at the head of the sentence in
the process of translation probably referrs to
− 39 −
those plural persons described beforehand.
Consequently the candidate hitobito (people
in general) should be rejected. Notice that in
example (8), the object of the preceeding sentence
is in fact these boys and the judgement of the
choice rule above was correct.
In case where there is no plural noun with
human attribute h , but there is some plural
noun without human attribute h , Japanese
candidate sorera (non-human plural) should be
adopted as shown in the following example.
Ken: Nancy, do you have any pets ?
健「ナンシー,あなたはペットを飼っています
か?」
Ken Nanshii, anata wa petto wo katte-imasu ka ?
Nancy: Yes, I do. I have two little birds. I
like them very much. ......... (9)
ナンシー「はい,飼っています。私は 2 羽の小さ
な鳥を飼っています.私はそれら (それらの鳥)
がたいへん好きです.
」
Nanshii Hai, katte-i-masu. Watashi wa
ni wa no chiisa-na tori wo katte-i-masu.
Watashi wa sorera( sorera no tori )ga taihen
suki-desu.
The choice rule applied to them ' in the
example above is;
22;*そ れ ら;2;E0=them;JA=12;SO=xX;SO<>h;T1<>u;
TA<>xXF;TZ<>h; J330;
Our translation system notifies pets as the
object of the preceeding sentence, and keeps
information about pets presented in the
preceeding sentence.
In some cases, we can select sorera (nonhuman plural) as the adequate Japanese
translation of they by taking account of the fact
that certain kinds of adjectives can not modify
human beings. See the following example.
− 40 −
Ken: Are they all yellow ? ...... (10)
selects sorera for the Japanese translation of
They in (11).
Different from these two they s above, in
case of them in example (12) the object of the
preceeding sentence ( students ) has attribute
h ( = human ) and FxX (plural) and the word
preceeding them is neither eat', sell' nor spell'.
Confirming all these conditions satisfied, the
following choice rule
minna kiiroi desu ka?
2;* 彼 ;2;E0=them;E-1<>eat;E-1<>sell;E-1<>*spell;J1<>#;EA<>,;SZ=h;SZ=FxX;VZ=u; I935;
Tom: Most students go to school by bus.
Look at the school buses.
トム「たいていの生徒はバスで学校に行きます.
スクールバスを見なさい.
」
Tomu Taitei-no seito wa basu de gakkou ni
iki-masu. Sukuuru-basu wo mi-nasai.
健「それら (それらのスクールバス)はみんな黄
色いですか?」
Ken Sorera( sorera no sukuuru-basu )wa
Tom: Yes. They pick up students near their
houses. ....... (11)
トム「はい.それら (それらのスクールバス)は
彼等の家の近くで生徒を拾い上げます.
」
Tomu Hai. Sorera( sorera no sukuuru-basu )
wa karera no ie no chikaku de seito wo hiroiage-masu.
Ken: Oh, good for them. ....... (12)
健「彼等 にとっては好いですね.」
Ken Karera ni totte-wa yoi desu ne.
selects the Japanese candidate beginning with
the character 彼 , namely 彼等 (karera) and 彼
女達 (kanojo-tachi). In this rule, the verbs which
do not usually supposed to take human beings
as their object and which appear in the first year
grade English textbooks are listed up. In general
more of the same kind of verbs should be listed
in the choice rule, or rather, we might define a
new attribute symbol for verbs which do not
usually take human beings as their object.
6 .Replace the translation of the substitutive
We see that they ' in examle ( 10 ) above is
modified by adjective yellow ' , and human
beings are usually not modified by yellow ' .
Thus the following choice rule
22;* そ れ ら ;2;E0=they;E1=all;E2=yellow;J3=?;OZ=x
;OZ<>h; J670;
selects sorera (non-human plural) from among
the Japanese candidates (explicitly, from
students and school buses in this case).
In the succeeding sentence ( 11 ), They is
situated at the head of the sentence, and the
Japanese translation of the preceeding sentence is
seen to be an interrogative one with expressions
sorera (...) ... desu ka? Consequently They
in ( 11 ) is considered to reffer sorera (...) in
the preceeding sentence. Verifying all these
conditions the following choice rule
22;* そ れ ら ;2;E0=they;FT;J1=3;E-1=.;E-2=yes;DJ-3=
「 ;OZ=x;OZ<>h;PJS=* そ れ ら( ;PJS=* で す か?」;
J715;
verb do by that of the verb in the preceding sentence (in case of status verbs).
In questions and answers in English
conversation, the verb appeared in the question
sentence is not repeated in the reply sentence
but auxiliary verb do is substitutively used
as we explained in the fourth article [2] of this
series. There we quoted the following example
sentences.
Ken: Do you practice judo every day?
健「あなたは毎日柔道を練習します か?」
Ken「Anata wa mai-nichi juudou wo renshuushi-masu ka?」
Mukami: Yes, I do.
......................... (1)*
Mukami「はい,練習します 。」
Mukami「 Hai, renshuu-shi-masu. 」
The Japanese translation of the verb of the
preceeding sentence is kept in the variable vr,
and the Japanese translation of do is replaced
Dependence on Context in case of English-Japanese
Machine Translation I−9 (K. Shibata)
by the value of vr. What we should be careful
with here is that in registering the Japanese
translation in variable vr, the progressive form
...te-iru of Japanese status verbs are modified
to infinitive form ... suru . Henceforth when
the translation system rewrites the Japanese for
do by that of the status verb of the preceeding
sentence, that expression should be remodified
back to te-iru form. In the following generation
rule, this remodification is executed by processing
command #SINKO;
3;6024;X;0;LT;VZ=?;PJS=*ていますか?;T0<>N;T3
<>p;DJ3<>出 来 る;E3<>may;J2#Z;J3#Z;J3#+VR$;J3#
SINKO;J3#-格;J3#DES; 9950;
In examples (8) and (9) of the preceeding section,
this generation rule is applied to change the
Japanese for do into shitte-i-masu and katte-imasu . Examples of such status verbs appearing
in the first year grade English textbooks are
shitte-iru (shiru), motte-iru (motsu), katte-iru
(kau) and so on.
7 . How to translate questions and answers
starting from an interrogation using interrogative pronoun which (part two).
In section 1 of the present article, we treated
an interrogative sentence with which at the
head of the sentence which contains in itself the
data for judging whether which modifies two
objects or not less than three objects. It is not
so in many other cases, though. We requote the
last part of the example sentences presented in
section 5 and continue the succeeding sentences.
Nancy: Do you know these boys, too?
ナンシー「あなたはこれらの少年も知っています
か?」
Nanshii Anata wa korera no shounen mo
shitte-i-masu ka ?
Ken: Yes, I do. They are Masao and Osamu.
............ (8)
健「はい,知っています.彼等は正夫(正雄,正
男)と治(修)です.」
Ken Hai, shitte-i-masu. Karera wa Masao to
− 41 −
Osamu desu.
Nancy: Which boy is Masao? ........... (13)
ナンシー「正夫(正雄, 正男)はどちら の少年で
すか?」
Nanshii Masao wa dochira no shounen desu
ka ?
Ken: This is Masao. I know him very well.
........... (14)
健「こちらが 正夫(正雄, 正男)です.私は彼を
とてもよく知っています.
」
Ken Kochira ga Masao desu. Watashi wa
kare wo totemo yoku shitte-i-masu.
In order to determine the Japanese for which' in
(13) above, the following choice rule is applied.
55;* ど ち ら ;2;E0=which;FT;J1=1;T1=h;T2=v;SZ=F;
SZ=h;SZ=K;SZ<>xX;BS=*They are ; g150;
This rule checks that which in question is
placed at the head of a sentece (FT), that it is
followed by a noun with human attribute h , and
that be verb (T2=v; ) succeeds. Next the rule
verifies that the attribute set SZ of the subject of
the preceeding sentence contains h (human),
F (conjunction and ), K (proper noun), but
no plural nouns ( xX ). It also confirms that the
preceeding sentence begins with They are ... .
All these conditions satisfied, the rule adopts
dochira for the Japanese translation of which
in (13). Of course, there remains the possibility
that the preceeding sentence is like They are
Takeshi, Masao and Osamu. So the choice rule
above should have verified that the preceeding
sentence does not contain commas (BS <>*,;).
As the Japanese for which in (13) has been
determied to be dochira , the Japanese for this
in ( 14 ) should be chosen to be kochira . In
the first article [1] of this series, we explained
that in case of conversations on a telephone,
English sentence like This is Kenji speaking.
should be translated into Japanese as Kochira
wa Kenji desu.
This time we have just
presented a second usage of kochira for this
; in the replying sentence to a question using
interrogative pronoun which . This is described
− 42 −
in the following choice rule.
22;* こ ち ら ;2;E0=this;T1=v;J2=18;T2=hM;BS=*Whic
h ;PJS=どちら; K220;
We stated in section 2 that the postpostion
which indicates that the preceeding noun is
the subject noun or subject pronoun in a reply
sentence to an interrogative sentence with
which at its head is ga instead of wa . In
example (14) above, the following generating rule
similar to the one quoted in section 2 is applied
to add ga to the tail of the subject pronoun
kochira here.
22;* あ な た 達 ;2;E0=you;TA=qr;EZ=then;BS=*We
don't ; K400;
This rule checks that the preceeding sentence
(BS) contains the expression We don't .... , and
that the present sentence in process of translation
is consequently the responding sentence with
then at the tail and with an interrogative ( q )
or a relative ( r ) preceeding you . With all
these conditions satisfied, the rule judges that
you in the present sentence corresponds to we
in the preceeding sentence.
9 .Selection of Japanese for then among sou
DJ2<>「 ;DJ3<>「 ;DJ4<>「 ;E2<>and;T1<>w;J0#+ が :
ga;J1#VR;J1#DES; S940;
sure-ba , sono toki , sorekara , sore-dewa , suruto and sono koro ni-wa in case
of inquiry sentence following a negative
statement.
8 .A second example where you is interpret-
In the preceeding article [5] of this series, we
have presented the following example sentences.
9;23;X;0;FT;LT;DJ0=こちら;EB=Which ;J-2<>0?!;
ed as plural.
In the sixth article [ 4 ] of this series, we
presented a choice rule which selects the plural
interpretation of you in case the subject of the
preceeding sentence has human attribute h
and is expressed in an expression xxxx, and
I . This time we examine the following example
appearing in Lesson 9 of a first year grade
English textbook.
Tom: Ken, look. We don't eat lunch in the
classroom.
トム「健,見なさい.私達は教室では昼食を食べ
ません.
」
Tomu Ken, mi-nasai. Watashitachi wa kyoushitsu de-wa chuushoku wo tabe-masen.
Tom: Oh, your new watch is pretty.
トム「ああ,あなたの新しい時計はきれいです.」
Tomu Aa, anata no atarashii tokei wa kirei
desu.
Kumi: Thanks. But it isn't mine.
久美「ありがとう.しかしそれ(その新しい時計)
は私のものではありません.
」
Kumi Arigatou. Shikashi sore( sono atarashii tokei )wa watashi no mono de-wa ari-
masen.
Tom: Then, whose watch is it? ............. (6)*
トム「それでは ,それ(その新しい時計)は誰の
時計ですか?」
Tomu Sore dewa, sore( sono atarashii tokei )
wa dare no tokei desu ka ?
Ken: Where do you eat lunch then ? ......... (15)
健「それではあなた達 はどこで昼食を食べます
か?」
Ken Sore dewa anatatachi wa doko de
chuushoku wo tabe-masu ka ?
The following choice rule selects anatatachi'
(you plural) for you' in (15) above.
We explained there that if the preceeding
sentence contains ... isn't then the then' at
the head of the responding sentence of inquiry
should be translated as sore dewa . In the
present article, we encountered then' in example
(15) in section 8. In that case, the preceeding
sentence contains ... don't and then' in (15)
is at the tail (LT) of an interrogative sentence
Dependence on Context in case of English-Japanese
Machine Translation I−9 (K. Shibata)
with an interrogative at the head (TA=q;). So the
context is very similar to that explained in the
preceeding article. We have the following choice
rule which applies to then' in (15).
66;*それでは;2;LT;E0=then;J1=?;TA=q;BS=* don't ;
m340;
10.Add the Japanese translation of the main
verb of the preceding sentence to that of a
prepositional phrase.
Look at the following example sentences.
Kumi: When does your school begin?
久美「あなたの(あなた方の)学校はいつ始まり
ますか?」
Kumi Anata no( anata-gata no )gakkou wa
itsu hajimari-masu ka ?
Tom: In September. It ends in June. ....... (16)
トム「 9 月に始まります.それは 6 月に終わりま
す.」
Tomu Ku-gatsu ni hajimari-masu. Sore wa
roku-gatsu ni owari-masu.
Independent prepositional phrase In
September in (16) is a reply to the preceeding
sentence which is an interrogative sentence
beginning with interrogative adverb When ' .
The following generation rule for the Japanese
translation of the prepositional phrase above
4;71;X;1;E0=in;FT;LT;T0=O;T1=t;T1=O;E2=.;EB=
When;BS=*?;VZ=?;J0#+VR$;J0#DES; B772;
adds the Japanese translation of the verb of the
preceeding sentence (J0#+VR$;). This makes the
Japanese translation more natural as a Japanese
expression. Notice that the generation rule
above verifies that the noun dominated by the
preposition begins with an uppercas letter (T0=O;)
and that the noun has time attribute t before
adding the translation of the verb to that of the
prepositional phrase.
− 43 −
References
[1] K a t s u y u k i S H I B A T A : D e p e n d e n c e o n
Context in case of English-Japanese Machine
Translation I-1. Fukuoka University Science
Reports, vol.35 No.1. pp.75-79, 2005
http://www 1 .rsp.fukuoka-u.ac.jp/chosho/
Cntxt1-1.html
[ 2 ] Katsuyuki SHIBATA: Dependence on Context
in case of English-Japanese Machine
Translation I-4. Fukuoka University Science
Reports, vol.36 No.2. pp.83-86, 2006
http://www 1 .rsp.fukuoka-u.ac.jp/chosho/
Cntxt1-4.html
[ 3 ] Katsuyuki SHIBATA: Dependence on Context
in case of English-Japanese Machine
Translation I-5. Fukuoka University Science
Reports, vol.37 No.1. pp.93-103, 2007
http://www 1 .rsp.fukuoka-u.ac.jp/chosho/
Cntxt1-5.html
[ 4 ] Katsuyuki SHIBATA: Dependence on Context
in case of English-Japanese Machine
Translation I-6. Fukuoka University Science
Reports, vol.37 No.2. pp.77-81, 2007
http://www 1 .rsp.fukuoka-u.ac.jp/chosho/
Cntxt1-6.html
[ 5 ] Katsuyuki SHIBATA: Dependence on Context
in case of English-Japanese Machine
Translation I-8. Fukuoka University Science
Reports, vol.38 No.2. pp.63-68, 2008
http://www 1 .rsp.fukuoka-u.ac.jp/chosho/
Cntxt1-8.html