WORD-QUEST MAIL (April, 2004)

    Context Dependency in Case of English-Japanese Machine Translation (I-7)
                        (English summary) 
                       
 In this article we investigate the examples of dependence on context found in 
Lesson 7 of the first grade five English textbooks.


1. Should "those" be translated as "arera no" or "sorera no" (Part 1)

 In the 5th article [2] of this series,  we investigated  the criterion for 
deciding the Japanese translation for "that" between "ano" and "sono". As its 
plural form version, we need the criteria for deciding the Japanese translation 
of "those" between "arera no" and "sorera no". In the example sentence in [2], 
the noun following "that" was different from eithr the subject or object of the 
preceding sentence. Consequently "ano" was chosen. In this article, we present 
a similar example in section 1,  and another example  for which "arera no" is 
rejected and "sorera no" is adopted by the converse reason in section 5 below.

Kenji: Who are those girls?   ............. (1)
 (Kenji: Arera no shoujo wa dare desu ka?)

Sam: They are my friends. We go to the same school.     ........ (2)
 (Sam: Kanojo-tachi wa watashi no yuujin desu. Watashitachi wa onaji gakkou 
       ni iki-masu.)

 The choice rule applied to decide the Japanese for "those" in (1) above is;

88;それらの;2;E0=those;E1<>people;DJ1<>SJ$;DJ1<>OJ$;DJ1<>SJ2$;DJ1<>OJ2$;E1<>*BS;
E1<>*BS2;D1<>回; d340;

This rule verifies that the noun directly following "those" is different from 
either the subject or object of the preceding sentence or of the sentence before
 the last,  and that  the noun  is contained neither in the preceding English 
 sentence (BS) nor in the sentence before the last (BS2). Consequently the noun 
doesn't refer to something already mentioned  but presents something new.  By 
these reasons the rule rejects "sorera no" and thus adopts "arera no".


2. Example of translating "they" into "kanojo-tachi" (the second case)

 In the previous article [3] of this series, we showed the example where "they" 
was translated into "kanojo-tachi", but the translated word was annihilated in 
the process of connecting the translated words. This time we present an example 
in wich "kanojo-tachi" is preserved in the final result of the translation.
 For "they" in the example (2) of the previous section, the choice rule is;

22;*彼女;2;SZ=W;SZ=x;OZ<>x;EA<>,; D990;

This rule checks that the attribute set of the subject of the previous sentence 
SZ contains 'W' (woman) and 'x' (plural symbol 's' at the tail of the word) and 
that the attribute set of the object of the preceding sentence OZ doesn't contain
 'x', it selects "kanojo-tachi" as the Japanese for "they". In fact, the subject 
of the preceding sentence is "those girls".


3. Example of translating "Mrs." as "fujin"

 For the title "Mrs.",  three Japanese words "san", "fujin" and "sensei"  are 
registered.  When "Mrs. AA"  appears with "Mr. AA"  in pair,  "Mrs. AA" will be 
translated as "AA Fujin".

Taro: How do you do, Mr Hill?
 (Taro "Hajime mashite, Hiru san.")

Becky's Father: Nice to meet you, Taro.
 (Bekii no Otou-san "Anata ni aete ureshii-desu, Taro.")

Taro: How do you do, Mrs Hill?  ........ (3)
 (Taro "Hajime mashite, Hiru Fujin.")

 The choice rule for "Mrs." in (3) above is;

88;*夫人;2;E0=Mrs;E-1=How do you do ,;T1=K;E1=*BS2;BS2=*How do you do, Mr; d130;

This rule requires that the preceding English sentence (BS) is "How do you do, 
(someone)?", that the attribute of the noun directly after Mrs. is 'K' (proper 
noun), that the proper noun is contained in the sentence before the last (BS2) 
and that the English sentence before the last contains "How do you do, Mr. (
someone)?" When all these conditions are satisfied, the rule selects 'fujin' as 
the adequate Japanese for "Mrs."
Remark: "Hajime mashite, Hiru Fujin." is, in fact, not so natural as a greeting 
by a junior highschool student.  The most natural translation would simply be 
"Hajime mashite", neglecting the English part "Mrs. Hill".  But here we do not 
require the translation to that extent.


4. Decision of what "too" refers to (a case of pronoun object)

 In the 4th article [1] of this series, we investigated whether ", too" refers 
to the object or to the subject of the sentence in case ", too" follows a noun 
object of a transitive verb. When the object of the verb is a pronoun, we are 
unable to apply similar criteria. Let's look at the following example. We remark 
that among the five English textbooks, only New Crown and New Crown (New Edition)
 write as  "too"  at the end of a sentence  not as ", too"  as in the other 
textbooks. Both expressions seem to be used in USA and the Great Britain.

Tom: Yes, I can. They are ki, hayashi and mori.
 (Tomu: Hai, deki-masu. sorera (sorera no tango) wa ki, hayashi to mori desu.)

Kumi: Right. They mean 'tree', 'woods' and 'forest'. Do you know this too? 
                                                            .............. (4)
 (Kumi: Sono-toori desu. Sorera wa 'ki', 'hayashi' to 'mori' wo imi-shi-masu.
        Anata wa kore-mo shitte-i-masu ka?)

 In the process of connecting the translated Japanese words from bottom up, 
the composition grammar rule to connect "wo shitte-iru (ru 7)", "kore" and 
"mo mata" into "kore-mo shitte-iru (ru 7)" is;

5;326;3;1;LT;J3=?;E2=too;E2<>* ;E-1=you;BS=*Yes, I can.;J0#-格;J0#VR;J1#+も;
J2#Z; 9662;

This rule reqires as preconditions that this part is at the end of a sentence 
(LT = LasT) followed by the interrogation mark '?' (and thus a question sentence)
 and that a preceding English word (E-1) which seems to be the subject is "you". 
So we here need to choose between "Anata-mo ...wo ...shi-masu ka?" and "Anata wa 
....-mo ...shi-masu ka?" And the rule conclude to choose "Anata wa ....-mo ...
shi-masu ka?",  if the preceding English sentence contains "Yes, I can."  This 
implies that the subject of the preceding sentence is "I", and that the subject 
of the present sentece "you" refers to the same person. By the principle that 
the Japanese postpostion "mo" (corresponding to English "too")  will not be 
attached to the already mentioned (pro)noun, "mo" is attached to the object, not 
to the subject.


5. Should "those" be translated as "arera no" or "sorera no" (Part 2)

 In section 1, we showed an example of selecting "arera no" for "those" since 
the noun following "those" is different either from the subject or from the 
object. In this section, we are going to show a contrary example of rejecting 
"arera no" and thus adopting "sorera no" by virtue of the preceding sentence 
information.

Tom: English has many riddles like this.
 (Tomu: Eigo ni-wa kono youna ooku-no nazo ga ari-masu.)

Kumi: Good. I like those word games. .......... (5)
 (Kumi: Sore wa yoi desu ne. Watashi wa sorera no kotoba-asobi ga suki desu.)

 The choice rule for "those" applied to (5) above is;

88;*それらの;2;E0=those;E-1=like;BS=* many ;BS=* like this.; d337;

The "word games" refered to "those" does not explicitly appear in the preceding 
sentence ("riddles" is paraphrased), but the preceding English sentence (BS) 
contains "... many ..." and "... like this", and so the noun phrase succeeding 
"those" is considered to point out what is mentioned in the preceding sentence, 
and by consequence the rule chooses "sorera-no" for "those".


6. The choice for "good" among "yoi", "yoroshii" and "zen"

 In our system dictionary, as the translations for "good", three Japanese words, 
"yoi" (adjective), "yoroshii" (interjection) and "zen" (abstract noun) are 
registered. If one replies "Good," in a conversation, we tend to select the 
interjection "yoroshii" for "good", but there are some cases where it is better 
to select "yoi" regarding the phrase as the abbreviation of "It's good." Such is 
the case of example (5)  in the previous section.  For "good" at the head of 
example (5),  the following two choice rules  are successively applied and is 
finally determined as "yoi";

5X;*好い;2;E0=good;FT;LT;DJ-1=「;E2=I;E3=like;BS=* has many ;BS=* like this.; 
V860;          Xよろしい ; 5好い
15;*好い;2;E0=good;FT;LT;DJ-1=「;E2=I;E3=like;BS=* has many ;BS=* like this.; 
9380;          5好い ; 1善 

 These rules check that "good" in question is a single word sentence (FT;LT) at 
the head of a speech (DJ-1=「;) followed by a sentence "I like ...". Namely the 
situation is; "Good. I like ..." Further they verify that the preceding English 
sentence (BS) contain "... has many ..." and "... like this." "Good. I like ..." 
is a reply to this. Therefore the speech seems to mean "That's nice. I like ..., 
too," and the rules select "yoi" for "good".
 And during the process of generating the Japanese translation, "sore wa" and 
"desu-ne"  are attached to respectively  from left and  from right,  and the 
translation for "Good." becomes "Sore wa yoi desu-ne."

1;5;X;0;FT;LT;E0=good;DJ-1=「;E1=.;PJS=があります。」;J0#+ですね;J0#それは+; 
1427;

 By the way,  this rule requires as one of the precondtion  that the preceding 
Japanese translation (PJS) contains "ga ari-masu.」" at the end of the preceding 
speech. In replay to such an utterance, "Good. ..." at the head of the answer 
must be translated into Japanese not simply by replacing the word registered in 
the dictionary but by further adding "sore wa" and "desu-ne" before and after 
the registered word to make the Japanese translation more natural.


7. Choice between "oboete-iru" and "omoi-dasu" for "remember"

 For English verb "remember", we have two Japanese "oboete-iru" (keep in memory) 
and "omoi-dasu" (recover the memory of). They are used as follows;

Kumi: What day is it today?
 (Kumi: Kyou wa nan-youbi desu ka?)

Tom: It's Friday. This afternoon our class collects cans. 
Do you remember?     ..................... (6)
 (Tomu: Kin-youbi desu. Kyou no gogo watashitachi no kurasu wa kan wo 
        atsume-masu. Anata wa oboete-i-masu ka?)

Kumi: Now I do. Thanks.  ..................(7)
 (Kumi: Ima omoi-dashi-mashita. Arigatou.)

 The choice rule for "remember" in (6) above is;

33;*をおぼえている;2;E0=remember;J1=?;E-1=you;E-2=do;E-3=.; M230;

This rule verifies  that the sentence under translation  includes "Dou you 
remember?" following dirctly after some statement (E-3=.;). If this condition 
is satisfied, the question mentioned above is translated as "Anata wa oboete-i-
masu ka?". This rule doesn't investigate the context inheriting the preceding 
sentences.
 But, what is interesting is the answer (5) to the question. The auxiliary verb 
"do" refers to the verb "remember" in the preceding sentence (4). In English 
that's all.  But in Japanese,  "remember" in the question (4)  and "do" (= 
"remember") in the answer (5) have different meanings and they should be 
expressed by different verbs. So we can't replace "do" by the value "oboete-iru" 
of the variable VR  of the verb of the preceding sentence.  Therefore the 
generation grammar rule applied in the process of translation of (5) checks that 
what is going to be translated is the answer to the question "Do you remember?", 
and then rewrites the Japanese for "do" as "omoi-dashi-mashita" rejecting the 
verb of the preceding sentence "oboete-iru";

3;24;X;0;LT;E0=I;E1=do;E2=.;E-1=now;DJ-2=「;BS=Do you remember?;J0#Z;J1#Z;
J1#+思い出しました; 6217;


8. Is the "It" in "It is date/time." a formal subject or refers to something?

 The "It" of "It's Friday." at the head of example (6) in the previous section 
is usually viewed as a formal subject and it's Japanese translation "sore" will 
be annihilated during the process of generating the translation. The grammar 
rule connecting "sore"(it) + "dearu"(be) + "Kin-youbi"(Friday) => "Kin-youbi 
desu" is

3;231;X;1;FT;E0=it;T1=v;T2=t;E2<>* on;E3<>that;PJS<>*の日が来ました。;J0#Z;
J1#DES; 6910;

This rule requires as a precondition that the preceding Japanese translation 
(PJS) does not contain "... no hi ga ki-mashita." If the preceding translated 
sentence contains such expression, it may happen that we translate "It is date/
time." as "Sore (= it) wa date/time desu."


9. Should "now" be translated as "ima" or "saa"?

 English word "now" is used as meaning "present time" (adverb, noun) as well as 
"Well, .." (interjection) for promption or starting a new subject. In case of 
"now" in example (7) in the two sections back, the system applies the following 
choice rule with very explicit conditions;

6X;*今;2;E0=now;FT;E1=I;E2=do;E3=.;DJ-1=「;BS=Do you remember?; Z937;
     6今 ; Xさあ 

This rule checks that the preceding English sentence (BS) is equal to "Do you 
remember?" and that the present sentence is "Now I do. ...". When the conditions 
are satisfied, this rule select "Ima" for the "Now" at the head. The conditions 
above are so concrete since the criterion for such "now" strongly depends on the 
lexical particularity  of the verb "remember"  in the preceding sentence  which 
would hardly be generalized.


                              << References >>

[3] Katsuyuki SHIBATA:  Context Dependency in case of English-Japanese Machine 
    Translation I-4.
   http://www1.rsp.fukuoka-u.ac.jp/chosho/Cntxt1-4.html

[3] Katsuyuki SHIBATA:  Context Dependency in case of English-Japanese Machine 
    Translation I-5.
   http://www1.rsp.fukuoka-u.ac.jp/chosho/Cntxt1-5.html

[3] Katsuyuki SHIBATA:  Context Dependency in case of English-Japanese Machine 
    Translation I-6.
   http://www1.rsp.fukuoka-u.ac.jp/kototoi/2004_4E.html#No290

             Michi-nari-ni (Along the road) (Part 2)
                        (English summary) 
                       
 To the previous issue "Wor-Quest mail No291 / Michi-nari-ni (Along the road)", 
many of my friends hurriedly responded me to provide me useful informatons.

 First, Mr. Uratani of NHK writes: I was surprised to know there is a Japanese 
person who have never heard of the expression "michi-nari-ni". I looked up 
Kodan-sha's Great Japanese Dictionary and found "michi-nari" as "according to 
the shape of the road". Sansei-do's Great Word Forest has a similar description.

 Yes. I verified that myself. But the first several Japanese dictionaries as 
Iwanami's Kojien, Gakuken's Great Japanese Dictionary, Sanseido's Shin Meikai 
Japanese Dictionary, Toyama Shobo's Great Word Ocean, Sanseido's Broad Word 
Forest, Shinchosha's Japanese Dictionary, Sanseido's Modern New Japanese 
Dictionary, The Great Japanese Dictionary (26 vol.) and Japanese Great 
Dictionary; first edition (vol. 20) (Rem: I wrote by mistake 28 vol. in the 
preceding article) have not contained the item "michi-nari(-ni)".  But now I 
undestand that there are(!) some Japanese dictionaries explaining about "michi-
nari(-ni)" other than those pointed by Uratani. Kadokawa's Japanese Dictionary 
and Shogakukan's Modern Japanese Example Dictionary' are two of such 
dictionaries.
 It is an interesting fact that Japanese Great Dictionary; first edition (vol.
  20) published in 1972 doesn't contain "michi-nari(-ni)", but the second 
edition of the same dictionary published in 2002 adopted the expression "michi-
-nari(-ni). It seems to me that the dictionaris including "michi-nari(-ni)" were
 published after about 1980, and I guess this would be confirmed by checking the
"Annual of the Modern New Japanese Words" for these ,say, thirty years. In other 
words,  I imagine that the expression "michi-nari(-ni)"  is a rather new 
Japanese idiom.

 Next Prof. Ogino of Tokyo Metropolitan University writes: Sanseido's Japanese 
Dictionary (third edition) has the item "michi-nari". The reason why few 
dictionaries have item "michi-nari" is that the expression can easily be 
understandable as "michi" + "nari".

 I don't agree with him at the latter part. Wthout looking up an adequate 
dictionary or be taught by an eruded person, one can not guess that "michi-nari
-ni" consists of "michi" + "nari (形 = shape, figure)" + "ni". Namely, "michi-
nari-ni" has an added meaning. When one says "michi-nari-ni", that "michi (road)"
 is winding and may have several crossroads, but neglecting the corssroads you 
must procede acoording to the curves of the road. But I found a dictionary which 
supports Ogino's suggestion. Taishukan's Meikyo Japanese Dictionary has no "michi-
nari" item, but its "nari" item explains the meaning of the word as "according 
to" and presents "michi-nari-ni yuku" as a pragmatic example. Prof. Kitada of 
Yokohama State University wrote to me a similar comment.

 Prof. Yano of Saitama University writes: I was born and bred in Kyoto, and I 
had never heard of the expression "michi-nari-ni" until when I first heard it 
in Tokyo or might have heard it spoken by someone from Tokyo while I was a 
university student. I remember having felt it was rather a strange expression.
I thought the expression was rather a modern idiom, but if your wife is familiar 
with the phrase, it might be a dialect of particular regions.

 I must apologize  that I misunderstood  what she stated  and gave a wrong 
description in the last issue of Word-Quest Mail. She comes from Yamaguchi, the 
most western part of the Mainland Japan where old Japanese expressions still 
remains to be used to some extent, but she says that the people there do not use 
the expression "michi-nari-ni". Whe she came up to the region north to Tokyo for 
work,  she first heard the expression  and didn't understand what it means and 
asked the explanation. That was her only experience of encountering "michi-nari
-ni". So Prof. Yano's guess may be right. I also feel that "michi-nari-ni" is 
rather a modern expression. I searched the expression in the volume of Muromachi 
Period of the Old Japanese Dictionary of Each Era, but couldn't find it.

 ex-Prof. haruta writes: In Shogakukan's Japanese Dialect Great Dictionary (the 
 latter volume p.1766, 1985), there is "nari" item as a suffix explaining; (3) 
direction. People in Tottori say "Go east 'nari-ni', turn to south 'nari-ni', 
and then turn right 'nari-ni' and go left 'nari-ni'".  Besides, (1) connecting 
to an action word, exprsses that the right after the action, (2) following a 
noun or a noun-connection form of a verb, implies to continue the situation or 
to include it. (4) connecting to a human or animal noun, implies "to the 
direction of". If your wife from Yamaguchi is familiar with the expression, it 
might be a dialect of the extreme west of Japan. At least, we don't use the 
expression in Hokuriku (the area along the Japan Sea).

 As for what I described about my wife's statement in the previous issue, I 
rectify as I did above in reply to Prof. Yano's comment. I also consulted 
Japanese Dialect Great Dictionary and noticed that the formal noun 'nari' or 
suffix 'nari' is used in various respective particular meanings in many narrow 
regions in Japan.

 I also asked about the expression to my faculty collegues who are born and 
bred in Fukuoka. Some of them had never heard of that expression. Others think 
"michi-nari(-ni)" is a standard (Tokyo) Japanese idiom and they understand what 
it means but they don't use it among themselves.

 To sum up, "michi-nari(-ni) is not commonly used in Kyushu, Yamaguchi, Hokuriku 
nor Kyoto.  If it happens to be used in those regions,  it must be used 
intentionally as a standard Japanese idiom by the influence of school education 
or of TV program. It is mainly used in Kanto (the area around Tokyo) and a part 
of Tohoku (the extreme north of the Mainland Japan), but it is a rather modern 
expression and I guess it would be difficult to find its pragmatic example 
before 1970.

             Michi-nari-ni (Along the road)
                   (English summary) 
                       
 While I was doing sightseeing in Italy last month, the guide in Milano and the 
one in Florence, both Japanese living there, said " ... michi-nari-ni yuku to ..
."  I have never heard of that Japanese expression  although I could imagine it 
would mean " ... going along the road ...". For my wife, born in Yamaguchi, the 
west extreme part  of the Mainland of Japan,  it was a common expression. As a 
standard Japanese (Tokyo) speaker,  I was curious about the expression,  and 
looked up verious Japanese dictionaries after coming back to Japan. I couldn't 
find the expression even in the Great Japanese Dictiobary (28 volumes) in the 
University Library. I first thought that the expression would be a dialect of 
Western Japan,  but It semms not. Looking up "michi-nari-ni" on the Internet 
Google search engine, I found 44300 sites, including a sightseeing guidance near
 Tokyo region  or a guide to a church  in the north of Japan.  So the search 
results show that the expression is used everywhere in Japan and very frequently.
 Then why isn't it on the Japanese dictionaries? The most similar example which 
I found in a dictionary was "yumi-nari-ni" (in the shape of a bow = yumi), but 
it is not so close, either. If someone knows the etymology of "michi-nari-ni", 
please tell me.

      Context Dependency in Case of English-Japanese Machine Translation (I-6)
                         (English summary) 
                       
 This time we treat Lesson 6 of the first year grade 5 textbooks by 3 publishers.

 It is often said  that mathematics is a step by step subject  and  that it is 
consequently impossible to continue learning math if one failed to understand 
the basics. It seems however that junior high English is really a step by step 
subject. The candidate choice rules and the translation generating grammar rules
 used in Lesson 1 are also used in Lesson 2 and thereafter, the rules used in 
Lesson 1 and 2 are also used frequently in Lesson 3 and thereafter, and so on. 

 In this paper, we only treat the phenomena which appear firstly in Lesson 6.


1. The Choice of the Japanese words for 'they' and 'their'

She is an English teacher.
 (Kanojo wa Eigo no sensei desu.)

We like her class.
 (Watashi-tachi wa kankjo no jugyou (kurasu) ga suki-desu.)

Are Ken and Kumi in the same class?
 (Ken to Kumi wa doukyuusei desu ka?)

Yes, they are. ................... (1)
 (Hai, sou desu.)

Who is their homeroom teacher?   .................(2)
 (Karera no hoomuruumu no sensei wa dare desu ka?)

Miss Kato is.   .......... (3)
 (Katou sensei desu.)

 As explained in the preceding article, our US System has 'karera', 'kanojo-
tachi', 'hito-bito' and 'sorera'  as the translations for 'they'.  We have 
hitherto seen the examples that adopt 'sorera', 'hito-bito' and 'karera'. This 
time we are going to show an exaple  in which 'kanojo-tachi' is adopted.  The 
choice rule for 'they' in (1) above is

22;*彼女;2;SZ=W;SZ=F;OZ<>x; E000;

and it verifies  that the attribute set SZ  of the subject  in the preceding 
sentence contains 'W' (Woman) and 'F' (conjunctive 'and' or 'or') and that the 
attribute set OZ of the object in the preceding sentence doesn't contain 'x' 
(plural form),  and consequently choose 'kanojo-tachi'.  In this example, the 
subject of the preceding sentence is "Ken and Kumi" and "Kumi" has attribute 'W'
 while 'and' has attribute 'F'.  We might have adopted  'karera'  for a group 
containing both man and woman  like "Ken and Kumi",  but our system takes the 
"feminist" view point of choosing 'kanojo-tachi'. Nevertheless this 'kanojo-
tachi' is deleted by a grammar rule during the process of the translation, and 
the final result becomes "Hai, sou-desu."

 As the Japanese translation of 'they' varies as above, so do those of 'them' 
and 'their'. In example (2) following (1) above, the choice rule applied to 
decide the Japanese for 'their' is;

88;*u;1;E0=their;E1<>impact;SO=h;SO=FxX; d620;

 This rule checks that the word following 'their' is not 'impact', and that the 
attribute set SO of the subject and the object of the preceding sentence contains
'h' (human) as well as 'F', 'x' or 'X' (each of the three implies 'plural'), and 
concludes  to select the candidates  with attribute 'u' (human').  Before the 
application of this choice rule,  the candidate 'kanojo-tachi no'  is already 
abandoned by another rule,  and consequently  the translation for 'their'  is 
determined to be 'karera no'.  It may be a formal contradiction that , while 
'they' in (1) and 'their' in (2) both refer to same "Ken and Kumi", 'they' in 
(1) is translated as 'kanojo-tachi'  and 'their' in (2) as 'karera no'.  Our 
system does not require the extreme rigidity whether to call 'kanojo-tachi' or 
'karera' the human group consisting both male and female persons. This fits the 
linguistic reality of human communications.


2. Example of translating the title 'Mr.', 'Mrs.' or 'Miss' 
   as 'Sensei' (= teacher)

 We pointed out in the third article [1] of this series that in the "classroom 
English" described in junior highschool textbooks, the title 'Mr.', 'Mrs.' or 
'Miss' of a teacher is translated in 'Sensei' (= teacher) in Japanese. This time
 we exhibit an example where 'Miss' is translated as 'Sensei'.
 
 For the sentence (3)  of the preceding section,  the Japanese for 'Miss'  is
 selected by the following choice rule;

88;*先生;2;T0=M;FT;J3=0;OZ=e;SJ2=先生; d700;

This rule applies to the words with attribute 'M' (= Mr., Mrs., Dr. etc) which 
are at the head of a sentence (FT = FirsT).  When the attribute set OZ of the 
object of the preceding sentence contains 'e' (education or culture)  and the 
Japanese translation SJ2  of the subject of the sentence before the last  is 
'sensei', the rule selects Japanese 'sensei' for the word in question. In fact, 
the values of SJ and SJ2 are respectively 'Ken to Kumi' and 'sensei'. We remark 
that in case there is no subject noun (i.e. pronoun subject) in the preceding 
sentence  or in the sentence before the last,  the former value of SJ or SJ2 
remains as it is.


3. Whether "there are/is" should be translated as "iru" or "aru"

 In some first grade junior highschool English textbooks, sentences containing 
"there are/is" begin to appear.  When there are/is (a) human being(s)  or 
animal(s), we say "ga iru" in Japanese, while the existence of inanimate(s) or 
plant(s) is expessed as "ga aru". In the usual answer sentence to an inquiry 
sentence, the subject is not mentioned in Japanese. Therefore we are obliged to 
refer to the preceding question sentence. Let's have a look at an example.

B: Are there many clubs in your school?
 (B: Anata no gakkou ni ooku-no kurabu ga ari-masu ka?)

T: Yes, there are. Ellen, Roy, and I belong to the International Club.
                                                 ................... (4)
 (T: Hai, ari-masu. Eren, Roi, soshite watashi wa kokusai-kurabu ni haitte-i-
 masu.)

B: What do you do in the International Club?    .............. (5)
 (B: Kokusai-kurabu de anata-tachi wa nani wo shi-masu ka?)

 In the example (4) above, the choice rule for "there are" is as follows;

33;u;1;T0=E;FT;SZ<>ah;JZ<>1;TZ<>ah; H060;

This choice rule applies  to the words (phrases) with attribute 'E' (= there 
are/is/exist/stand) which are at the head of a sentence (FT = FirsT). When 
the attribute SZ of the subject of the preceding sentence doesn't contain 'a' 
(animal) nor 'h' (human), and if there is no word with attribute 'a' or 'h' 
in the successive part of the sentence, the candidate with attribute 'u' (human)
is rejected. This is equivalent to reject "ga iru". At this stage, the recorded 
informations of the preceding subjects and their attributes are;

sj = 1クラブ  sz = ?xxeByつ     sj2 = 1サッカー部, sz2 = emyつ
      (club)                           (soccer club)

and consequently, no human or animal noun is recorded.


4. An example where 'you' is interpreted as plural

 In example (5) of the preceding section,  "what do you do"  is registered to 
have three Japanese translations, i.e. "anata wa nani wo shi-masu ka", "Anata 
wa nani wo shi-te-i-masu ka" and "Anata-tachi wa nani wo shi-masu ka".  The 
choice rule applied to (5) to select one of them is;

XX;*あなた達は;2;E0=what do you do;J-1=|;BS=*, and I ;SZ=h;SZ=F; e772;

This rule verifies that the preceding English sentence (BS) contains ", and I ",
 and that the attribute set SZ of the subject of the preceding sentence contains
 'h' (human)  as well as 'F' (conjunction 'and' or 'or'),  and consequently 
selects the plural form "anata-tachi".


5. Whether 'city' should be translated as "toshi" or "machi"

 The English common noun 'city' is usually translated as "toshi" in Japanese, 
but there is also a case where "city of Edo" is translated as "Edo no machi". 
If the 'city' in the preceding sentence is translated as "machi" according to 
the context, the 'city' in the next sentence has the possibility of also beeing 
translated as "machi".  Therefore  in a choice rule  which select "toshi" for 
'city', we insert the condition that "the Japanese translation of the preceding 
sentence doesn't contain "machi".

Here is an example;

Everyone is very busy.
 (min-na ga totemo isogashii desu.)

It's an interesting city. .................. (6)
 (Sore wa omoshiroi toshi desu.)

The choice rule applied to 'city' in (6) above is the following.

11;街;2;E0=city;E1<>of;E-1<>this;EA<>part;TA<>g;PJS<>街; 0825;

 This rule rejects the candidate "machi"  after verifying several conditions 
among which the preceding Japanese translation (PJS) not containing "machi".


6. Whether the verb 'watch' should be translated 
   as "wo miru" or "no ban wo suru"

 In our system dictionary, 'watch' as a verb has two Japanese "wo miru" (look 
at) and "no ban wo suru" (stand guard over).

Alice watches Humpty.  ............... (7)
 (Arisu wa Hanputii wo mi-masu.)

The choice rule applied to 'watch' in  (7) above is;

33;*を見る;2;E0=watch;T0<>pGP;J1=1;T1=h;T1=K;J2=0;VZ<>ijk; O522;

 If someone 'watches' someone (proper noun T1=h;T1=K;),  and if the attribute 
set VZ of the verb of the preceding sentence contains neither 'i' (transitive 
verb), 'j' (intransitive verb) nor 'k' (double transitive verb), then the rule 
above choose "wo miru" as the adequate translation. Is is of course possible 
that someone may stand guard over someone, but in such a case, there should 
precede some sentences explaining the situation, and consequently the attribute 
set VZ contains 'i', 'j' or 'k'. If there is no such explanations and someone is 
said to 'watch' someone,  the choice rule above  judges that "wo miru"  has a 
greater possibility than "no ban wo suru".

 The example sentence (7) above is cited from "Alice in the Wonderland", and it 
has naturally the preceding context, but our US System cancels the values of 
all the variables  concerning the preceding contexts  when a new chapter or 
section begins. The sentence (7) is placed in such a position and hence there 
was no context informations.


                                 << Reference>>
[3] Katsuyuki SHIBATA:  Context Dependency in case of English-Japanese Machine 
    Translation I-3.
   http://www1.rsp.fukuoka-u.ac.jp/chosho/Cntxt1-3.html