1. Nishigauchi 2000[1]

advertisement
Maki Irie / Nishigauchi summary
1. Nishigauchi 20001 'Scrambling and reconstruction at LF'
Nishigauchi's claim can be summarized as follows:
(1)
The 'scrambled' NPs are reconstructed not only to their original position but also to the
periphery of various projections (edge of CP, vP, and so on) that dominate their original
position.
And any restrictions as to which projection(s) that dominate their original position?
Is it assumed that the OS order in Japanese is uniformly derived by the movement of the O from its
theta position to its surface position? Is the movement assumed to obey the Subjeacency? Is the
movement assumed to be 'cyclic' adjunction?
Specifically, his claim implies that there are potentially 3 (or more, depending on the number of the
edge positions) positions in which the scrambled NP can be interpreted:
(2)
(3)
Surface: NPDL ....[ .... ec .... ] (ec= the position which NP DL is thematically related.)
LF:
a. NPDL ....[ .... ec .... ] -- no reconstruction
b. ... [ NPDL [ .... ec .... ] -- the edge of the intermediate phrase
c. ... [ .... NPDL .... ] -- the original position (reconstruction)
The point of this paper is (3b): reconstruction to the intermediate phrase is possible. His arguments
for his claim are based on binding, specifically condition C and condition A. His arguments are
constructed on the basis of binding theory (Condition C and A) in conjunction with scope interaction
1) between a QP and scope-bearing X0? elements (SBs); and 2) between a wh-phrase and the question
marker Q.
For the first argument on the basis of SB, Nishigauchi cites Fox's 1999 and 2000 generalization
in (N33), and makes the predictions in (N34):
(N33)
(N32)
(N34)a.
b.
(N35)
(N36)
Scope reconstruction feeds Condition C
Scope reconstruction should be impossible in the structural configuration in (32).
[QP ... r-expression1 ...]2 ... pronoun1 ... t2
In (35), the QP must take scope over the scope-bearing element SB.
In (36), the QP need not take scope over the scope-bearing element SB.
[QP ... r-expression1 ...]2 ... pronoun1 ... SB ... t2
[QP ... r-expression1 ...]2 ... SB ... pronoun1 ... t2
So, we should know (more) about:
--Condition C (D) effects in Japanese
-- Fox's 1999 and 2000 generalization in (N33), and makes the predictions in (N34)
--The scope interaction between a QP and a scope-bearing X0
The example (N26) is presented to show that the prediction in (N34a) is confirmed:
(N26)
[Yamada-senseii-no kurasu-no nan-nin-no gakusei-o]j kanozyoi-ga tj
Yamada-teacher-gen class-gen
how many
students-acc
she-nom
MIT-e ik-ase tagatte iru no?
MIT-to
go-make want
is
Q
1
The published version of this paper is in Gengo Kenkyu 2002, but the version I have is dated November 14,
2000. It seems that he makes only this version available on the web. I assume that the claims and the arguments
presented in the 2002 published version are not radically different from the one in 2000, however, I think I need
to make sure that when I cite it in my paper.
1/5 Summary of Nishigauchi
Irie
Maki Irie / Nishigauchi summary
'How many students of Prof. Yamadai's class does shei want to send to MIT?'
(many > want) *(want > many)
Nisigauchi's observation is that (N26) is unambiguous, in that many must take scope over want. In
other words, the example has only the meaning that Prof. Yamada has some specific group of students
in mind and she wants to hire them, and the speaker is asking the number of the students of that
specific group (i.e., wide scope reading of many; referential reading). It does not mean that the
speaker does not presuppose that Prof. Yamada has some students in mind and asks the number Prof.
Yamada's goal for recruitment (i.e., narrow scope of many; nonreferential reading). [This is exactly
what is predicted, as the QP containing the R-expression and how many must be reconstructed below
the pronoun, to avoid a violation of Condition C2.<==Redo. A bit unclear as to what is modified by "
to avoid a violation of Condition C."]
What we need to know (more) about now includes:
--specific and non-specific readings out of "#-classifier-gen-N"
[Although (N34a) is not directly related to the intermediate reconstruction site <==So, I guess
Nishigauchi is taking (N26) as an instance of Eg* and (N31) as its corresponding Eg.], he presents the
example in (N31) on the basis of (N34b), claiming that it is ambiguous, in that how many can take
scope over or below the scope-bearing element omotta
(N31)3
[Yamada-senseii-no kurasu-no nan-nin-no gakusei-o]j Taroo-ga
Yamada-teacher-gen class-gen
how many
students-acc
Taro-nom
[kanozyoi-ga tj MIT-e ik-ase tagatte iru to]
omotta
no?
she-nom
thought
Q
MIT-to
go-make want
is that
'How many students of Prof. Yamadai's class did Taro think shei wanted to send to MIT?
(many > thought; thought > many)
Nishigauchi's argument proceeds as follows. No reconstruction takes place to get the reading many >
thought. To get the reading thought > many, however, the dislocated NP must be reconstructed
within the embedded clause. However, it cannot be reconstructed to its original position, due to
Condition C. Thus, the following structure should be possible at LF, with reconstruction in the edge
position (in this case, Spec-CP):
(N62)
[how many]j does Taro think (=SB) [CP [tj students of Prof. Yamadai's class]k shei wants to
send tk to MIT]
2
I believe the following example is expected to be ambiguous, which was made by myself on the basis of N26,
taking the condition C factor away (EgN26):
(i) [Yamada-senseii-no kurasu-no nan-nin-no gakusei-o]j gakubutyook-ga tj
Yamada-teacher-gen class-gen how many students-acc
dean-nom
MIT-e ik-ase tagatte iru no?
MIT-to go-make want
is Q
'How many students of Prof. Yamadai's class does the deank want to send to MIT?'
expected: (many > want) (want > many)
3
I have still not had a clear distinction among the possible readings, but my understanding on the possible
interpretations are the following: See my remarks below (N36) above.
(i) how many > think > want (There is x number of students the speaker presupposes that Taro thinks that
Prof. Yamada wants to send these students to MIT, and the speaker is asking x)
(ii) think > how many > want) (Taro has in mind a certain number of students Prof. Yamada wants to
send to MIT, and the speaker is asking the number of the students Taro has in mind. --??)
(iii) think > want > how many (Neither the speaker nor Taro has any specific students in mind, and the
speaker is asking how many Taro thinks Prof Yamada wants to hire --??)
2/5 Summary of Nishigauchi
Irie
Maki Irie / Nishigauchi summary
And is it not possible to analyze (N62) (well, ignoring the 'separation of "how many" from the DL) as
an instance of the Major Object construction?
The second argument on SB is on the basis of Condition A4. He extends Fox's generalization to
the case of Condition A. Given the claim in (1) and Condition A, he makes the following (positive)
prediction as in (4), and presents the example (N46):
(4)
(N38)
(N47)
(N46)
the QP can take scope over or below SB in the following configuration.
(Based on the first paragraph of 5.2)
[QP ... anaphor1 ...]2 ... r-expression1 ... SB ... t2
[QP ... pronounr1 ...]2 ... r-expression1 ... SB ... t2
[zibun-zisini-no kurasu-no nan-nin-no gakusei-o]j Yamada-senseii-ga tj
kanozyoi-no
self/she-gen
class-gen
MIT-e ik-ase
how many
students-acc Yamada-teacher-nom
tagatte iru no?
MIT-to go-make want
is
Q
'How many students of heri class does Prof. Yamadai want to send to MIT?'
(many > want) (want > many)
According to Nishigauchi's observation, the example is ambiguous, in that how many can take scope
over or below the SB want. Namely, both the reading that asks the number of a specific group Prof.
Yamada wants to send to MIT (referential) and the one that asks her goal for recruitment to send
students to MIT (nonreferential) are possible. The crucial point is that the referential reading (wide
scope reading of how many) is possible. The dislocated NP containing the QP how many and the
reflexive zibun-zisin must be below the R-expression Yamada-sensei, due to the condition A,
nevertheless, the dislocated NP must be above want in order to get the wide-scope reading. Hence,
the following structure should be available at LF:
(5)
[how many]j does Prof. Yamadai [vP [tj students of zibun-zisini's class]k want to send tk to MIT]
(Based on Nishigauchi (51b))
He further discusses the wh-licensing cases, and argues that the wh-phrase must be
reconstructed, but the reconstruction site for the wh-phrase does not have to be its original position
and can be the edge position, assuming that the wh-phrase must be licensed within the scope of the
question marker in Japanese.
(N72)
[Hanakoi-no dono syasin-o]j Masao-ga [kanozyoi-ga tj itiban ki-ni-itte iru ka]
Hanako-gen
which picture-acc Masao-nom she-nom
siri-tagatte
iru koto
want-to-know
is
best
fond-of
be Q
fact
'[which picture of Hanakoi]j Masao wants to know [[shei likes tj best]Q]'
For the dislocated wh-phrase to be licensed, it must be within the embedded clause headed by Q.
However, a Condition C violation incurs if it is reconstructed to its original position. Hence, the whphrase must be reconstructed to the periphery position:
(N74)
Masao wants to know [[which picture of Hanako i]j shei likes tj best ] Q]
In addition to Condition C, he continues on the wh-phrase case with Condition A. He observes
that zibun-zisin is ambiguous in the following example; namely it can refer to the matrix subject
Masao or the embedded subject Hanako:
4
Although he put the examples containing a pronoun kanozyo, I think that the example containing zibunzisin is
more crucial in the sense that it constitutes the evidence for reconstruction in the intermediate position.
3/5 Summary of Nishigauchi
Irie
Maki Irie / Nishigauchi summary
(N79)
[Zibun-zisini,j-no dono syasin-o]k Masaoi-ga [Hanakoji-ga tk itiban ki-ni-itte iru ka]
self-gen
which picture-acc Masao-nom she-nom
siri-tagatte
iru koto
want-to-know
is
best
fond-of
be Q
fact
'[which picture of selfi,j]k Masaoi wants to know [[Hanakoj likes tk best]Q]'
In the above example, the crucial reading is that Masao wants to know which pictures of
his(=Masao's) Hanako likes best. To get this reading, the dislocated wh-phrase cannot be
reconstructed to its original position, due to Condition A. However, it must be reconstructed to the
position within Q, as the wh-phrase must be licensed. Hence, the following representation should be
possible at LF:
(N80)a. Masaoi wants to know [[which picture of selfi]k Hanakoj likes tk best]Q] zibun-zisin=Masao
2. My plan
After I read this paper, I thought that his claim may not be on the right track. The reasons why I
felt so are the following:
(6)
a. Use of zibun-zisin, as an anaphor
b. Use of Condition C effects
c. my unclear judgments of crucial examples
First, Condition A seems to be crucial for his arguments, however, I do not think zibun-zisin is
an anaphor. [Second, I do not feel the following example perfect:
(7)
Taro-ga Taro-no tomodati-o butta.
I think the above example should be unacceptable due to Condition C, as the second occurrence
of Taro is c-commanded by the first occurrence of Taro. <==There is some error here?]
In addition, the following example is getting not unacceptable to me, although it does not sound
terrific:
(8)
(A and B are talking about gossips of auto companies, and they started talking about Toyota,
then)
Asoko-ga [mukasikara Toyota-to torihiki siteita kouzyou]-o uttaeta n datte sa.
Third, I do not have a clear understanding of scope interaction he uses. I tried, and once I felt
that I could get the interpretations he refers to as non-referential and referential (and use of kanozyo
referring to Yamada-sensei, which I'm not sure if it is pragmatically natural), then I can get both
interpretations in (N26), which Nishigauchi claims has the referential reading only.
(N26)
[Yamada-senseii-no kurasu-no nan-nin-no gakusei-o]j kanozyoi-ga tj
Yamada-teacher-gen class-gen
how many
students-acc
she-nom
MIT-e ik-ase tagatte iru no?
MIT-to
go-make want
is
Q
'How many students of Prof. Yamadai's class does shei want to send to MIT?'
(many > want) *(want > many)
I would like to argue against Nishigauchi's claim. Specifically, my claim would be as follows:
(9) a. Intermediate reconstruction (3b) is not possible.
<==How do you think you could show that?
b. Nishigauchi's Hsub's (condition C and condition A) are not right in Japanese.
I am not sure if Nishigauchi's Hmain (i.e., (1)) would be falsifiable. He only makes positive predictions.
I think there is only one negative prediction he makes, (N34a), however, this is not directly related to
4/5 Summary of Nishigauchi
Irie
Maki Irie / Nishigauchi summary
the intermediate reconstruction site, as far as I understand.
[However, if I could show that (9b) is on the right track and predictions on the basis of (9a) are
not disconfirmed, I think it would suggest that his claim be reevaluated. <==I think you conclusion
should be stated in stronger terms than this.]
What I have to understand can be summarized as follows:
(10) a.
Heycock 1995 and Fox 1999 and 2000 (for scope-bearing elements and interpretations
involving scope-bearing elements)
b. Hayashishita dissertation (for scope interaction) <==Since Ringe's work is on scope
interaction among XPs (well, I don't remember him discussing scope interaction between
XP and Y0 …), his thesis might not be directly relevant. But, of course, it would insofar as
you address the issues of what intuitions count as a manifestation of a wide scope
distributive reading based on LF c-command. So, it may well be relevant when you try to
articulate your experimental design.
c. Ueyama's system of scrambling <==By now, you must have a fairly good understanding of
this.
Here is what I think you should do.
You should think about which one(s) of the issues noted above by yourself or by me is/are the most
important (and hence you should work on them).
The answer may depend upon what you would like to claim in your paper.
So, you should prepare an outline of your paper, giving the organization of the paper, and stating for
each section what you want to (be able to) say. I.e., you should do the two things at the same time.
5/5 Summary of Nishigauchi
Irie
Download