1 日本語の寄生空所構文

 Parasitic Gap Constructions in Japanese
URL: http://www.tscc.tohoku-gakuin.ac.jp/~sacl/Abe-CV.htm
1. Introduction
(1) Which articles did John file t without reading e?
t = trace, e = parasitic gap (PG)
(2) a. Which articles did John file t without reading them?
b. *Which articles did John file them without reading t?
(3) He’s a man that [anyone who talks to e] usually likes t.
(4) a. He’s a man that [anyone who likes talking] usually likes t.
b. *He’s a man that [anyone who likes t] usually likes talking.
2. Typical Properties of Parasitic Gap (PG)
i) Wh-phrases in situ do not license PG:
(5) a. Which article did you file t [without reading e]?
b. *Who filed which article [without reading e]?
ii) The real gap cannot c-command PG (Anti-c-command requirement):
(6) *Who [t met you [before you recognized e]?
iii) Island effects on PG:
(7) This is the man John interviewed t before
:
a. expecting us to tell you to give the job to e.
b. *expecting us to ask you which job to give to e.
c. *reading the book you gave to e.
d. *expecting you to leave without meeting e.
(8) a. He’s a man that [anyone who talks to e] usually likes t.
b. He’s a man that [anyone who tells people to talk to e] usually likes t.
c. *He’s a man that [anyone who meets people who talk to e] usually likes t.
d. *He’s a man that [anyone who asks when to talk to e] usually likes t.
1
iv) Case matching effects on PG:
(9) a. It was John that Mary believed t to be a genius before Susan proved e to be (a
genius).
b.?*It was John that Mary believed t was a genius before Susan proved e to be (a
genius).
3. Apparent PG Constructions in Japanese
(10) B@#%&' t [e ;
] ?=
i) Wh-phrases in situ do license PG:
(11) %&'B@# [e ;
] ?=
ii) No anti-c-command requirement on PG:
(12) :
[%&'
e C!9]A+
iii) No island effects:
(13) a. B@#%&' t [e ;$)
]?=
b. B@#%&' t [e 57(9]?=
iv) No case matching effects:
(14) :#[4 e(-dat) (7]
t In these cases, e is identifiable as pro.
(15) a. John said that he would win the game.
b. “Have you seen John recently?” “Yes. I saw him in the library yesterday.”
(16) a. %&'[>
3/6"].
b. %&'[pro 3/6"].
(17) a.0,%&'#-
2>#1<85*-
b.pro 0,%&'#-
pro 1< pro 85*-
4. Genuine Cases of PG in Japanese
4.1. Reconstruction Effects with respect to Condition A
(18) a. Which books about himself did John file t before Mary read e?
b. *Which books about herself did John file t before Mary read e?
Reconstruction into the real gap is possible, but not into PG.
2
(19) !)* t [ e '&] #
Zibun must be interpreted as John, not Mary.
But this fact can be accounted for by simply assuming the pro strategy.
(20) a. Which picture of himself did [every boy who saw e] say Mary liked t?
b. *Which picture of herself did [every boy who saw e] say Mary liked t?
(Munn 1994, p. 407)
Reconstruction into PG is possible, but not into the real gap.
(21) !)"$[e %][ t (
]
(22) [e %][ t (
]
!)
"$
If e were pro, it would not allow Condition A reconstruction. Hence, (21) and
(22) manifest genuine instances of PG.
The case-matching effect should be observed in this case:
(23) ?*!)"$[e(-acc)
%][ t ]
(24) ?*[e(-acc)
%][ t ]
!)
"$
Even if e is taken as pro, (23) and (24) will induce weak crossover (WCO)
violations irrespective of which subject zibun refers to, since the fronted
wh-phrase involves long-distance movement.
4.2. Sloppy Identity
- Takahashi (2006):
(i) Japanese (apparent) PG constructions allow ‘sloppy’ reading, as shown in (25) and
(26) below, where “what picture of self” denotes Taroo’s picture and Hanako’s
picture at the same time.
(ii) But they are not real PG constructions, but rather involve ellipsis, just like such an
English sentence as (27) below.
3
(25) ['* e #)&]
t $)
!("%
(26) !("%['* e #)&]
t $
)
(27) Which picture of himself did John look at before Peter did?
(sloppy reading possible)
(28) Which picture of himself did John look at t before Peter threw out e?
(sloppy reading impossible)
Note: Even though adjunct PGs (e.g. (28))) do not allow the sloppy reading,
subject PGs (e.g. (29)) do.
(29) What kind of picture of himself did the person who ordered John to throw e
away insist that Bill keep t?
(sloppy reading possible)
The availability of ‘sloppy reading’ indicates that there is reconstruction to both
gaps.
Such a reading would not be available if one of them is pro.
(25) and (26) are genuine instances of PG constructions.
The case-matching effect should be observed in these cases:
(30) !("%['* e(-acc) #)&]
)
t (sloppy reading impossible)
(31) ['* e(-acc) #)&] t )!("%(sloppy reading impossible)
Under Takahashi’s (2006) claim that e involves NP-ellipsis, the unavailability of
the sloppy reading to (30) and (31) is unexpected.
References
Abe, Jun and Chizuru Nakao (2009) “On ATB-Movement and Parasitic Gaps in
Japanese,” In Sun-Woong Kim, ed., 2009 Visions of the Minimalist Program,
Proceedings of the 11th Seoul International Conference on Generative Grammar,
Hankuk Publishing Co., pp. 1-15.
Munn, Alan (1994) “A Minimalist Account of Reconstruction Asymmetries,”
Proceedings of the North East Linguistic Society 24, ed. by Merce Gonzalez, GLSA,
University of Massachusetts, Amherst, pp. 397-410.
Takahashi, Daiko (2006) “Apparent Parasitic Gaps and Null Arguments in Japanese,”
Journal of East Asian Linguistics 15: 1-35.
4