Upload
amanda-martinez
View
234
Download
0
Tags:
Embed Size (px)
DESCRIPTION
kjhkjh
Citation preview
Intervention effect in Korean wh-questions:
Indefinite and beyond
Young-Sik Choi *
Department of English Language and Literature, Soonchunhyang University, Asan,
Chungnam 336-745, South Korea
Received 15 March 2004; received in revised form 24 December 2006; accepted 7 January 2007
Available online 15 May 2007
Abstract
I claim that the intervention effect in Korean wh-questions can receive a natural account with the
proposed two ways of scope taking of in situ wh-words: in situ interpretation of indefinite wh-words via
unselective binding by the question morpheme versus movement of way why. Also on the basis of the
interesting contrast of the intervention effect in wh-questions with the two types of wh-words in various
constructions, for which Beck and Kim (Beck, S., Kim, S.-S., 1997. On wh- and operator scope in Korean.
Journal of East Asian Linguistics 6, 339384) cannot offer a principled account, I claim that the intervention
effect is not a movement but scope phenomenon as opposed to them.
# 2007 Elsevier B.V. All rights reserved.
Keywords: Indefinite; Question morpheme; Scope; Intervention; Restriction
1. Introduction
Various approaches have beenmadewithin syntactic theories for the analysis of in situ wh-word
scope taking.Among them are the LFwh-movement approach byHuang (1982), and the LFwh-in-
situ interpretation approach byBaker (1970), an ideawhich has been adoptedwithin theminimalist
framework in the form of either unselective binding or absorption. However, a careful examination
of the nature of in situ wh-words in Korean, which is a typical wh-in-situ language, suggests two
ways of scope taking: in situ interpretation of indefinite wh-words via unselective binding by the
question morpheme, and movement of way why. It will be shown that scope taking of wh-words
along this way can offer a more satisfactory account for the intricate pattern of the intervention
effect in Korean wh-questions.
www.elsevier.com/locate/lingua
Lingua 117 (2007) 20552076
* Tel.: +82 41 530 1124; fax: +82 41 530 1491.
E-mail address: [email protected].
0024-3841/$ see front matter # 2007 Elsevier B.V. All rights reserved.doi:10.1016/j.lingua.2007.01.001
Beck and Kim (1997) observes that wh-questions with a negative polarity item (NPI,
henceforth) followed by a wh-word in Korean exhibit the intervention effect while with a
reversed word order, the effect does not arise, as illustrated in (1) and (2) (also see Sells, 2001;
Sohn, 1995; Suh, 1990, for related discussions).1,2
(1) a. *Amwuto nwukwu-lul chotayhaci an hayss-ni?anyone who-ACC invite NOT did-QM
Who did no one invite?
b. *Amwuto Mary-lul eti-se mannaci an hayss-ni?anyone M-ACC where-at meet NOT did-QM
Where did no one meet Mary?
c. *Amwuto Mary-lul encey mannaci an hayss-ni?anyone M-ACC when meet NOT did-QM
When did no one meet Mary?
d. *Amwuto Mary-eytayhay ettehkey malhaci an hayss-ni?anyone M-about how talk NOT did-QM
How was no one talking about Mary?
(2) a. Nwu-ka amwuto chotayhaci an hayss-ni?who-NOM anyone invite NOT did-QM
Who invited no one?
b. John-i eti-se amwuto mannaci an hayss-ni?J-NOM where-at anyone meet NOT did-QM
Where did John meet no one?
c. John-i encey amwuto mannaci an hayss-ni?J-NOM when anyone meet NOT did-QM
When did John meet no one?
d. John-i ettehkey Mary-eytayhay amwu-ekey-to malhaci an hayss-ni?J-NOM how M-about anyone-to talk NOT did-QM
How was John talking to no one about Mary?
Y.-S. Choi / Lingua 117 (2007) 205520762056
1 The intervention effect should be understood as referring to the effect in wh-questions, since sentences with a wh-
word other than way why can also be construed as a yesno question in Korean.2 Some native speakers find the sentences in (1) are not that bad (Nam-kil Kim, personal communication). I suggest that
it has to do with the pragmatic discourse-linking effect of the wh-words in (1) since they can be typically discourse-linked
like which-phrases in English as shown in (i):
(i) Speaker A: Malun salam-kwa ttungttunghan salam-i oko issta.
thin man-and fat man-NOM coming
There are a thin man and a fat man coming.
Speaker B: Ne-nun nwukwu-lul te coaha-ni?you-TOP who-ACC more like-QM
Which one do you like better?
One can check this discourse-linking factor by asking whether the hearer can answer the question in (1a), for example, by
giving an answer like someone you dont know. This answer is virtually impossible even to native speakers who report
(1a) is more or less acceptable, while one can give this answer to a similar English question below in (ii) (Barry Schein,
personal communication).
(ii) Who did no one invite?
Once the discourse-linking effect is properly controlled, I believe one can give a consistent judgment on the examples in
(1).
Interestingly, as observed by Cho (1998) and Choi (2002), the intervention effect in (1) does
not arise when the NPI is followed by way why as in (3).3
(3) Amwuto way Mary-lul chotayhaci an hayss-ni?anyone why M-ACC invite NOT did-QM
Why did no one invite Mary?
Thus, the paradigm above in (13) poses a nontrivial problem to Beck and Kims (1997)
observation: Why is it that the intervention effect does not arise in (3) in contrast to (1)? In this
paper, I will offer a formal account for the contrast in the intervention effect in (1) and (3). I will
also show that the present account for the intervention effect can extend to other wh-questions
involving double object constructions and embedded wh-question constructions too, where the
intervention effect does not arise quite against Beck and Kims (1997) expectation.
The organization of the presentation is as follows: in section 2, I briefly review Beck and
Kims (1997) proposal. I will show that they cannot offer a principled account for the interesting
contrast in the intervention effect as exhibited by the wh-questions with way why and with the
other wh-words as in (1) and (3). This suggests that the main premise of Beck and Kim (1997) for
the intervention effect as a movement phenomenon cannot be maintained. I examine the property
of so called wh-words in Korean in section 3 and claim that wh-words other than way why are
indefinites in the sense of Lewis (1975) and Heim (1982). In section 4, I provide an alternative
account for the contrast in the intervention effect in wh-questions in (1) and (3) with two ways of
scope taking of wh-words, namely, unselective binding and movement, which I suggest
essentially follows from the different status of wh-words as indefinites. In section 5, I will show
how the present proposal extends to account for the intricate pattern of intervention effects in
wh-questions involving double object constructions and embedded wh-questions, for which
Beck and Kim (1997) cannot offer a satisfactory account. Section 6 is the implication of the
present proposal for the intervention effect with various types of QPs. Section 7 is the conclusion.
2. Beck and Kims approach to intervention effect
According to Beck and Kim (1997), who assume LF wh-movement (Huang, 1982), the
ungrammaticality of (1) can be accounted for in terms of Minimal Negative Structure Constraint
which prohibits negation that triggers a Negation Induced Barrier from intervening between the
wh-operator and its variable at LF: the definitions of Negation Induced Barrier and Minimal
Negative Structure Constraint are given in (4) and (5), respectively.
(4) Negation Induced Barrier (Beck and Kim, 1997:347)
The first node that dominates a negative quantifier, its restriction, and its nuclear scope
is a Negation Induced Barrier (NIB).
(5) Minimal Negative Structure Constraint (Beck and Kim, 1997:347)
If an LF trace b is dominated by a NIB a, then the binder of b must also be dominatedby a.
Y.-S. Choi / Lingua 117 (2007) 20552076 2057
3 An anonymous reviewer notes that wh-questions with wh-word . . . NPI order are consistently preferred, irrespective
of the types of wh-words and irrespective of the positions they appear. Recent research, however, shows that way why
behaves differently from the other wh-words in that it does not show the intervention effect in wh-questions (see Cho,
1998; Choi, 2002, among others).
The examples in (1) will have the LF in (6) under Beck and Kims (1997) system, which
violates Minimal Negative Structure Constraint (MNSC, henceforth). This is because the VP,
which dominates the negation an not, and is the Negation Induced Barrier (NIB, henceforth)
intervenes between the wh-operator and its variable, given the assumption that the NPI should be
under the scope of negation (see Ladusaw, 1980).
(6) [CP wh-operatori [IP [VP an [VP NPI ti ]]]]
Beck and Kim (1997) assumes that the subject stays in its base position of Spec of VP at S
structure in Korean and that negation an not is adjoined to VP, adopting domination in terms of
nodes, but not categories.
When it comes to (3), MNSC, however, predicts it to be ungrammatical contrary to the fact,
since the VP that dominates the negation an not and is the NIB intervenes between the wh-
operator and its variable at LF in (7).
(7) [CP whyi [IP [VP an [VP NPI ti ]]]]
To the extent that (3) is acceptable, hence exhibiting no intervention effect, the main premise of
Beck andKim (1997) for the intervention effect as amovement phenomenon cannot bemaintained.
Then why is it that the wh-question in (3) with way why does not show the intervention effect in
contrast to the ones in (1) with the other wh-words? To answer this important question, we need a
careful examination of the nature of wh-words in Korean to lay the groundwork for our analysis
later on.
3. The nature of wh-words
The interpretation of wh-words in Korean type languages including Chinese and Japanese can
vary, depending on the context where they occur, and thus the nature of wh-words in these
languages has been a topic of much discussion (Aoun and Li, 1993; Cheng, 1991; Choi, 1937;
Chung, 1996; Kim, 1991; Kuroda, 1965; Li, 1992; Nishigauchi, 1990, among others). As for
Korean, Kim (1991) claims that wh-words are uniformly quantifiers and Chung (1996) claims
that they are uniformly wh-interrogatives, both suggesting that wh-words have their own
quantificational force.4
A close scrutiny reveals, however, that wh-words in Korean, with the exception of way why,
are indefinites in the sense of Lewis (1975) and Heim (1982). Before examining the nature of
wh-words, I will have a brief review of the indefinite as discussed in Heim (1982). Heim (1982),
following Lewis (1975), suggests that an indefinite exhibits quantificational variability without
its own inherent quantificational force, and immunity to an island. The following examples in (8)
illustrate the behavior of the indefinite as mentioned just above (Heim, 1982:123):
(8) a. If a man owns a donkey he always beats it.
As for every x, y, x an individual, y a donkey if x owns y, x beats y.
b. Sometimes, if a cat falls from the fifth floor, it survives.
As for some x, x a cat if x falls from the fifth floor, x survives.
Y.-S. Choi / Lingua 117 (2007) 205520762058
4 Chung (1996) and Kim (1991), however, invite the question of why the adjunct wh-word way why has a
wh-interrogative reading only, if wh-words are uniformly quantifiers or wh-interrogatives underlyingly from which
various readings derive.
The indefinites a man, a donkey and a cat in (8) can scope out of the island with their
quantificational force determined by adverbial quantifiers always, and sometimes at LF, which
are unselective binders as shown by the informal logical notations. With the two important
characteristics of an indefinite in mind, namely, quantificational variability and island immunity,
let us turn to the following adverbs of quantification constructions in (912) from Choi (2002),
which I believe provide an ideal test for the status of wh-words as indefinites:
(9) a. [CPNwui-ka o-myen] (proi) nul wuli-lul pangmwunhanta.who-NOM come-if always us-ACC visit
For every x, x an individual, if x comes, x visits us.
b. [CPJohni-i eti-lo oychwulha-myen] (proi) nul nusskey tolaonta.J-NOM where-to go out-if always late returns
For every x, x a place, if John goes out to x, he returns home late (from x).
c. [CPJohni-i encey o-myen] (proi) nul wuli-ul pangmwunhanta.J-NOM when come-if always us-ACC visit
For every x, x a time, if John comes at x, he visits us (at x).
d. [CPNay-ka ettehkey John-eytayhay malha-myen]I-NOM how J-about talk-if
Mary-nun nul hwalul naykonhanta.
M-TOP always get angry
For every x, x a manner, if I talk about John in x, Mary gets angry (with x).
(10) a. [CPNwui-ka o-myen] (proi) kakkum wuli-lul pangmwunhanta.who-NOM come-if sometimes us-ACC visit
For some x, x an individual, if x comes, x visits us.
b. [CPJohni-i encey o-myen] (proi) kakkum wuli-lul pangmwunhanta.J-NOM when come-if sometimes us-ACC visit
For some x, x a time, if John comes at x, John visits us (at x).
c. [CPJohni-i eti-lo oychwulha-myen] (proi) kakkum nusskey tolaonta.J-NOM where-to go out-if sometimes late returns
For some x, x a place, if John goes out to x, he returns home late (from x).
d. ?[CPNay-ka ettehkey John-eytayhay malha-myen]I-NOM how J-about talk-if
Mary-nun kakkum hwalul naykonhanta.
M-TOP sometimes get angry
For some x, x a manner if I talk about John in x, Mary gets angry (with x).
(11) *[CPJohni-i way o-myen] (proi) nul wuli-lul pangmwunhanta.J-NOM why come-if always us-ACC visit
For every x, x a reason, if John comes for x, he visits us (for x).
(12) *[CPJohni-i way o-myen] (proi) kakkum wuli-lul pangmwunhanta.J-NOM why come-if sometimes us-ACC visit
For some x, x a reason if John comes for x, he visits us (for x).
Choi (2002:3237, slightly modified)
As shown above in (9) and (10), the interpretation of wh-words as universal quantifiers or
existential quantifiers is determined by the adverbial quantifier nul always and kakkum
Y.-S. Choi / Lingua 117 (2007) 20552076 2059
sometimes in the main clause, which convey a universal quantificational force and an
existential quantificational force, respectively. The informal logical notations also indicate that
these wh-words can take scope out of an island. Meanwhile, way why cannot simply be
construed in an analogous way as shown in (11) and (12), which are simply ungrammatical. This
state of affairs crucially suggests that wh-words in (9) and (10) are indefinites exhibiting
quantificational variability and island immunity, whereas way why in (11) and (12) is not an
indefinite.5
One may thus reasonably believe that the different status of wh-words as indefinites also
affects their mode of construal as wh-interrogatives. One salient property of wh-questions in
Korean is that at the surface level, a wh-expression remains in situ with the question morpheme
(QM, henceforth) marking its scope, whereas in English the same expression is preposed into the
sentence-initial position for scope taking as shown in (13) and (14) (Cheng, 1991; Suh, 1987,
among others).
(13) a. Ne-nun [CPJohn-i nwukwu-lul mannassta-ko] sayngkakha-ni?you-TOP J-NOM who-ACC met-COMP think-QM
Who do you think John met?
b. Ne-nun [CPJohn-i way nussessta-ko] sayngkakha-ni?you-TOP J-NOM why late-COMP think-QM
Why do you think John came late?
(14) a. Who do you think John met?
b. Why do you think John came late?
Although standard Korean also has QMs ci, nya, (e)yo and (su)mnikka, all homophonous for
[+Q, +WH] and [+Q, WH] like ni, I will use ni as the representative one throughout, unlessotherwise necessary.6
Besides the wh-question interpretation, (13a), with the rising intonation at the end, is also
construed as a yesno question with the wh-word interpreted as someone, whereas the analogous
construal is simply impossible in (13b). One may interpret this as indicating that unlike way
why the interpretation of the indefinite wh-word in (13a) covaries with the choice of the QM,
exhibiting quantificational variability, given that QM ni in standard Korean is homophonous for
both [+Q, +WH] and [+Q, WH].In fact in Kyengsang dialects spoken in the southern part of Korea the interpretation of
wh-words other thanway whyvaries preciselywith the choice of a particular questionmorpheme,
which ismorphologically distinct for a yesno question and awh-question, unlike standardKorean.
Thus, with the yesno question morphemes ka and na, the indefinite wh-word yields an existential
Y.-S. Choi / Lingua 117 (2007) 205520762060
5 An anonymous reviewer points out that wh-words within the if-clause in (9) and (10), when combined with the
morpheme inka conveying the existential quantificational force, do not covary with the adverb in the matrix clause that
conveys the universal quantificational force as in (i).
(i) [CPNwu-inkai-ka o-myen] (proi) nul wuli-lul pangmwunhanta.who-INKA-NOM come-if always us-ACC visit
For some x, x an individual, if x comes, x always visits us.
This state of affairs suggests that there is something like the quantifier-indexing (operator-indexing) mechanism (Heim,
1982:146) at work in (i), prohibiting nonlocal unselective binding of an indefinitewh-word by the binder nul always into
the domain of the other binder inka in (i).6 Like ni, QMs ci and nya are used when the speaker is equal or superior to the hearer in social status, whereas (e)yo and
(su)mnikka are used when equal or inferior in social status to the hearer.
quantifier reading,whereaswith thewh-questionmorphemes ko and no, the samewh-word yields a
wh-interrogative reading as illustrated in (15) (see Suh, 1987:2ff).
(15) a. Nwu-ka wass-na?who-NOM came-QM
Did someone come?
b. Nwu-ka wass-no?who-NOM came-QM
Who came?
I thus suggest that the wh-interrogative reading of the indefinite wh-word in (13a) arises via
unselective binding of the indefinite wh-word by the QM with [+Q, +WH] at LF, forming an
operator variable construction, with the QM as the wh-operator marking the scope of the
indefinite wh-word that functions as a variable with no inherent wh-feature and also serves as
the restriction of the QM. Hence in (13a) with an indefinite wh-word, the QM that serves as the
wh-operator and the indefinitewh-word as its restriction appear separated at the surface level unlike
English in (14), thus informally translated as for which x, do you think John metman (x)?7 In an
analogous way, I suggest that the existential quantifier reading of the indefinite wh-word in (13a)
arises via unselective bindingof the indefinitewh-word by theQMwith [+Q,WH]atLF, yieldingthe reading informally represented as for some x, do you think John met man (x)?8,9
Y.-S. Choi / Lingua 117 (2007) 20552076 2061
7 An anonymous reviewer points out that althoughwh-words (other thanway why) can be used as indefinites as shown in
(9) and (10) wh-indefinites and non-wh-indefinites (like plural common nouns or expressions corresponding to a/an +N in
English) should differ in their feature composition and that the features distinguishing one from the other are responsible for
the (un)availability of a wh-interrogative reading. I agree with the reviewer in that wh-indefinites are different from other
indefinites, since the latter cannot have awh-interrogative reading evenwith the appropriate QM. I suggest the former have a
WH-feature, which however is underspecified in terms of [WH], borrowing the terminology from phonology. In otherwords, indefinite wh-words are not inherently wh-interrogatives or existential quantifiers, but require some other element,
namely, QM for the relevant readings (also see Aoun and Li, 1993; Cheng, 1991; Nishigauchi, 1990, among others).8 As an anonymous reviewer points out, unlike QM with [+Q, +WH], QM with [+Q, WH] can appear without a wh-
word. This is clearly shown in the Kyengsang dialect in (i) where the two types of QMs are morphologically distinct (see
Suh, 1987; Choe, 1995, among others).
(i) a. Ni-ka ka-na?
you-NOM go-QM
Are you going?
b. *Ni-ka ka-no?
you-NOM go-QM
Are you going?
Given this, one may argue that the existential quantificational reading of an indefinite wh-word may not be attributed to
the unselective binding by the QM with [+Q, WH]. However, it would be conceptually nice if the existentialquantificational reading is also via the same mechanism. Please note that (ia) with the QM with [+Q, WH] isconstrued as Are you going or not? whose structure can be roughly represented as the following with the QM serving as
the yes-no operator, and binding a sentential variable: [WH-either or]x you are going x (see Katz and Postal, 1964:104,
among others). The optionality of a wh-word is thus expected given the semantics of the QM with [+Q, WH].9 As one can see, the existential quantificational reading of the indefinite wh-word obtains in (ia), which, as an
anonymous reviewer points out, may indicate that QM with [+Q, WH] has nothing to do with the existentialquantificational reading of the indefinite wh-word.
(i) a. Nwu-ka wassta.who-NOM came
Someone came.
b. Nwukwu-inka-ka wassta.who-INKA-NOM came
Someone came.
Now, as for way why in (13b), given that it is not an indefinite and thus has its own inherent
wh-feature, I suggest that it should undergo movement into Spec of CP at LF, to form an operator
variable chain, checking off its wh-feature via spec-head agreement with the QM with
[+Q, +WH], thus without leaving a restriction in situ but a variable. (13b) therefore will be
translated informally as for which reason x, do you think John came late for x?10,11 Below it will
be shown that the difference in ways of scope taking between the two types of wh-words has
important implication for the intervention effect in Korean wh-questions.
4. The nature of intervention effect
Nowletus turn to the interestingcontrast in the interventioneffectbetween (1),with (1a) repeated
in (16) as a representative example, and (3) with the adjunct wh-wordway why, repeated as (17).
(16) *Amwuto nwukwu-lul chotayhaci an hayss-ni?anyone who-ACC invite NOT did-QM
Who did no one invite?
(17) Amwuto way Mary-lul chotayhaci an hayss-ni?anyone why M-ACC invite NOT did-QM
Why did no one invite Mary?
As discussed in section 2, the grammaticality of (17) suggests that the main premise of Beck
and Kim (1997) for the intervention effect as a movement phenomenon cannot be maintained,
since movement of way why into the operator position of Spec of CP at LF does not induce the
expected intervention effect, although negation intervenes between the wh-operator and its
variable at LF, leading to the violation of MNSC (also see Hoji, 1985; Tsai, 1994, among others,
for the intervention effect as a movement phenomenon, although expressed in different ways).
Moreover, if the present proposal is on the right track that the wh-word as in (16) is an indefinite,
which is interpreted in situ via unselective binding, the intervention effect in (16) should not
possibly arise from movement of the wh-word across the negation.
Y.-S. Choi / Lingua 117 (2007) 205520762062
Note, however, that the existential quantificational reading in (ia) does not necessarily refute the idea that the indefinite
wh-word is construed as an existential quantifier via unselective binding by the QM with [+Q, WH] in an interrogativesentence. With no external binder in (ia), namely, QM with [+Q, WH], the indefinite wh-word should have theexistential closure (Heim, 1982) introduced for its interpretation, which is the null morpheme corresponding to the overt
morpheme inka in (ib) that conveys the existential quantificational force.10 The proposed two ways of scope taking of in situ wh-words in Korean wh-questions fit into the minimalist thesis of
movement as a last resort in Chomsky (1995:200) who suggests that a shorter derivation is preferred to a longer one, and
if the derivation D converges without application of some operation, then that application is disallowed.11 As an anonymous reviewer notes,way why behaves differently from other wh-words with respect to the island effect
as illustrated in (i) (also see Chung, 1991, 1996).
(i) a. Ne-nun [nwu-ka ssun chayk-ul] ilkess-ni?you-TOP who-NOM wrote book-ACC read-QM
Who is the person x such that you read a book x wrote?
b. Ne-nun [os-ul ettehkey ipnun salam-ul] coaha-ni?you-TOP clothes-ACC how wear man-ACC like-QM
What is the manner x such that you like a man who gets dressed in x?
c. *Ne-nun [John-i way ssun chayk-ul] ilkess-ni?you-TOP J-NOM why wrote book-ACC read-QM
What is the reason x such that you read a book John wrote for x?
This follows straightforwardly under the present system: movement of way why versus in situ interpretation of indefinite
wh-words via unselective binding, with the former subject to the movement constraint of subjacency (Chomsky, 1973).
Then how can we account for the contrasting grammaticality of (16) and (17)? Given our
proposal that way why in (17), not being an indefinite, should move into Spec of CP at LF for
proper interpretation as a wh-interrogative by forming an operator variable chain, the adjunct
wh-word necessarily takes scope over the subject QP (NPI) in Spec of IP at LF, assuming quantifier
raising is driven to avoid a type mismatch a la Heim and Kratzer (1998). I will thus propose the
following constraint, assuming LF is the only level valid for the semantic interpretation including
scope following Chomsky (1995)12:
(18) A wh-interrogative should take scope over a QP in wh-questions at LF.
I therefore argue that the ungrammaticality of (16) is due to violation of the constraint in (18)
but not MNSC, as opposed to Beck and Kim (1997).
Now, the question one should ask is why (16) violates (18). According to our proposal,
the wh-word in (16) is an indefinite whosewh-interrogative reading arises via unselective binding
at LF, with the QM with [+Q, +WH] as the wh-operator binding the indefinite wh-word and
marking its scope. Hereafter, I mean by the QM the one with [+Q, +WH], unless otherwise
specified.
Thus, in our system given the nature of the QM as the wh-operator for an indefinite wh-word,
the position of the QM at LF bears crucial importance in determining the scope of a
wh-interrogative relative to a QP in wh-questions.
Now, I will first try to locate where the QM in Korean is base-generated, although it has been
widely assumed to project CP albeit without much discussion (Cheng, 1991; Joo, 1989;
Nishigauchi, 1990, among others).
As shown in (19), Korean verbal morphology including the QM is heavily agglutinating,
consisting of more than one bound morpheme, with the QM at the end determining the clause
type as a question.
(19) John-i nwukwu-lul manna-ss-ni?J-NOM who-ACC meet-PAST-QM
Who did John meet?
Given the recent strict projectionist hypothesis (Chomsky, 1993; Pollock, 1989, among others)
viewing the IP system as an extension of the VP system, with each inflectional morpheme of the
verb heading a separate functional projection within the system, the particular sequence of the
bound morphemes on the verb in (19) including the QM thus leads one to postulate the following
clausal structure in (20), where the past tense morpheme and the QM project TP and IP
respectively, right on top of VP:
(20) [IP [TP [VP V]T] I]
Y.-S. Choi / Lingua 117 (2007) 20552076 2063
12 The constraint in (18) may sound like a mere stipulation. However, it is also compatible with the independent claims
in the literature against a QP taking scope over a wh-phrase (Chierchia, 1991, 1993; Engdahl, 1986, among many others):
Quantifiers which can apparently take scope over a wh-phrase (assuming pair list answer signals it a la May, 1985, 1988)
are limited to universal QP types. Besides, the argument for pair list answer as signaling wide scope of a QP over a
wh-phrase is problematic, since wide scope of a QP over a wh-phrase should not be allowed, given that the semantics of
questions is sets of propositions (Karttunen, 1977) and that quantification is defined on propositions not sets of
propositions (see Chierchia, 1991:80ff).
Also, as originally claimed by Kim (1991), the following example where the QM nya fails to
show complementary distribution with the indirect speech marker ko that is standardly assumed
to head CP further suggests that QM heads IP:
(21) John-un [CPMary-ka mwues-ul sat-nya-ko] mwulessta.J-TOP M-NOM what-ACC bought-QM-COMP asked
John asked what Mary bought. (Kim, 1991:227, slightly adapted)
One may however still suggest that nya-ko in (21) is a reduced form of nya-ha-ko as in (22),
and that the verb ha say and the indirect speech marker ko head VP and CP, respectively, with
the QM nya heading its own CP (see Chung, 1996).
(22) John-un [CPMary-ka mwues-ul sat-nya-ha-ko] mwulessta.J-TOP M-NOM what-ACC bought-QM-say-COMP asked
John asked what Mary bought.
On this view, the lack of complementary distribution between the QM and the indirect speech
marker in (21) cannot necessarily be associated with the QM and the indirect speech marker
heading different projections of IP and CP, respectively. However, it should be noted that hako in
(22) as one unit is the direct speech marker as distinguished from ko that is the indirect speech
marker, although the two may be historically related (see King, 1994; Lee, 1993a; Martin, 1992;
Sohn, 1999; Song, 1996, among others).13 The grammaticality of (21) and (22) thus indicates that
the matrix predicate mwulessta asked can take CP headed by either of the two speech markers
and that QM nya in the embedded clause is compatible with both speech markers. Given the
status of hako as the direct speech marker heading CP, (22) thus cannot serve as an argument
against the proposal that QM heads IP.
One can find additional evidence for the QM heading IP below in (23). As already shown in
(21) and (22), although the matrix predicate mwulessta asked can take a clause headed by the
indirect speech marker ko as well as the direct speech marker hako, QMs ni, ci, and eyo in the
embedded clause can cooccur with only hako.
(23) a. *John-i [CPMary-ka wass-{ni/ci/eyo}-ko] mwulessta.
J-NOM M-NOM came-QM-COMP asked
John asked whether Mary came.
b. John-i [CPMary-ka wass-{ni/ci/eyo}-hako] mwulessta.
J-NOM M-NOM came-QM-COMP asked
John asked whether Mary came.
Now recall that it has been a wellknown observation in the literature that the choice of a
particular complementizer reflects certain properties of the verbal system of the clause, which is
formalized as theCI agreement rule. This is illustrated by the cooccurrence of the complementizer
that and the finite I, and the complementizer for and the nonfinite I in English below in (24).
(24) a. I think that John left.
b. Id prefer for John to do it.
(Chomsky and Lasnik, 1977:434)
Y.-S. Choi / Lingua 117 (2007) 205520762064
13 The indirect speech marker ko is first attested in the late 19th century whereas the use of the direct speech marker hako
goes as far back to the 15th century (see King, 1994; Martin, 1992).
With the above observation in the literature, (23) therefore can be construed as an instance of
CI agreement, with the QM in I subject to cooccurrence restriction imposed by the speech
marker in C, thus further supporting the proposal for QM heading IP. In the meantime, note that
QMs including those in (23a) can freely occur in the matrix clause without any restriction as
shown in (25), further suggesting that the restricted distribution of the QM with respect to a
particular speech marker in the embedded clause in (23) is an instance of CI agreement.
(25) Nwu-ka wass-{ni/ci/nya/sumnikka/eyo}?who-NOM came-QM
Who came?
At this point one remaining question is how selectional restriction by the matrix predicate
mwulessta asked canbe satisfied as in (21) and (22) if theQMis indeed in thehead of IP but notCP.
For this, I suggest that selectional restriction is satisfiedviaCI system followingGrimshaw (1986):
The matrix predicate selects the head of its CP complement, which in turn selects the head of IP.14
Finally, I would like to present the following as a further argument for the QM heading IP, an
argument as originally made by Whitman (1989). As one can see in (26), the suppositive
morpheme ci (SM, henceforth), expressing modality of supposition enters into complementary
distribution with the QM ni.
(26) a. Mary-nun kuttay yeppess-ci.
M-TOP that time beautiful-SM
Mary was probably beautiful at that time.
b. *Mary-nun kuttay yeppess-ni-ci.
M-TOP that time beautiful-QM-SM
This means that SM and QM are base-generated to compete for the same syntactic position. To
the extent that one claims that QM heads CP, one has to say that the SM expressing modality also
heads CP, a position which goes against the observation in the literature that modality is typically
expressed in INFL cross-linguistically (see Chomsky, 1955 and subsequent works; Pollock, 1989;
Y.-S. Choi / Lingua 117 (2007) 20552076 2065
14 Unlike English in (i), complementizers in Korean, that is, both direct speech marker hako and indirect speech marker
ko, can introduce both an interrogative clause and a declarative clause, as illustrated in (ii) with ko.
(i) a. *I wonder that John met Mary.
b. I wonder if John met Mary.
(ii) a. John-un [CPMary-ka nwukwu-lul mannass-nya-ko] mwulessta.J-TOP M-NOM who-ACC met-QM-COMP asked
John asked whom Mary met.
b. John-un [CPMary-ka Tom-ul mannass-ta-ko] sayngkakhanta.
J-TOP M-NOM T-ACC met-IND-COMP think
John thinks that Mary met Tom.
This suggests that complementizers in Korean are lacking their inherent [Q] feature, which may have to do with the factthat clause typing morphemes including the QM in Korean are residing in I, marking their respective clause types
(Jean-Roger Vergnaud, personal communication). As shown in (iii), sentence ending morphemes such as indicative marker
(IND) ta, imperative marker (IMP) la, and propositive marker (PROP) ca show complementary distribution with the QM as
illustrated with the QM ni. This indicates that they all head IP, competing for the same syntactic position with the QM.
(iii) a. *John-i Mary-lul mannass-ni-ta.
J-NOM M-ACC met-QM-IND
b. *John-i Mary-lul manna-ni-la.
J-NOM M-ACC meet-QM-IMP
c. *Wuli-ka Mary-lul manna-ni-ca.
we-NOM M-ACC meet-QM-PROP
Whitman, 1989, among others). Thus, the data in (2126), together with the particular sequence of
the verbal morphology including the QM as in (19) lead to the conclusion that QM heads IP but not
CP.15
Now, let us go back to (16). Under the VP internal subject hypothesis (Koopman and
Sportiche, 1991, among others), the surface subject will be in Spec of IP to satisfy the subject
requirement of a sentence, namely, the constraint of EPP, given that I, but not T is the head of a
sentence under the present proposal for the Korean clausal architecture in (20) (see Chomsky,
1981, 1995, among many others). With the present claim for the QM heading IP, the LF for (16)
will be (27), with the subject QP in Spec of IP under the fairly standard assumption that quantifier
raising is triggered to resolve a type mismatch (see Heim and Kratzer, 1998)16,17,18,19:
(27) [IP QPj [I QMi [TP [VP tj V whoi-ACC ]]]]
As one can see, (27) violates (18) since the QM as the wh-operator that unselectively binds the
indefinite wh-word and marks its scope at LF is below the scope of the subject QP in Spec of IP,
assuming c-command in the sense of Reinhart (1976) as a syntactic notion of scope.20 Thus, (16)
is correctly predicted as ungrammatical.
Y.-S. Choi / Lingua 117 (2007) 205520762066
15 As for the claim that QM heads IP, an anonymous reviewer wonders whether an embedded question in (i) ending with
the QM but not allowing the overt complementizer to be attached is a CP or IP.
(i) Na-nun [Mary-ka way wass-nunci-{*ko/*hako}] molunta.I-TOP M-NOM why came-QM-COMP do not know
I do not know why Mary came.
Given the proposal that selectional restriction of the matrix predicate is satisfied via CI in Korean, one may maintain that
there is an empty C as shown in (ii) with the matrix predicate selecting C, which in turn selects the QM nunci with [+Q,
+WH] in I, the latter being subject to cooccurrence restriction imposed by the former.
(ii) Na-nun [CP[IPMary-ka way wass-nunci] C] molunta.I-TOP M-NOM why came-QM do not know
I do not know why Mary came.
Also, as discussed in footnote 14, complementizers in Korean do not have the inherent [Q] feature. Hence, postulatingan empty C as in (ii) does not seem to be a problem.16 Although Korean is a strict head-final language, I will hereafter use a head-initial notation for the LF representations
for the readers convenience.17 One may find it hard to follow the claim that subject is higher than the morpheme that determines the clausal type as
an anonymous reviewer points out, especially since complementizers that and if as in English marking the [Q] force ofthe clause they introduce are typically higher than the subject of the clause they introduce. However, as discussed in
footnote 14, the fact that unlike English the same complementizer in Korean can introduce both an interrogative clause
and a declarative clause suggests that clause typing morphemes including the QM in Korean head IP, determining the
status of CP in terms of [Q].18 In a neutral context, the example in (i) with the order of the subject before temporal adverbial sounds the most natural
as compared with the ones with the subject either right after manner or temporal adverbial, although subtle. One may test
this in an embedded context, for example (to better control a potential intervening factor such as topicality) (Chungmin
Lee, personal communication). One may interpret this as indicating that the surface subject as in (i) is in a position higher
than Spec of TP, namely, Spec of IP, assuming temporal and manner adverbials are licensed by Tand V, being adjoined to
their respective maximal projections.
(i) John-i ecey yelsimhi kongpwuhayssta.
J-NOM yesterday hard studied
John studied hard yesterday.19 Since the precise position of negation is orthogonal to the present analysis, I will not notate it at the LF representations
throughout the paper.20 The definition of c-command in Reinhart (1976:32) is given in (i).
(i) Node A c(constituent)-commands node B if neither A nor B dominates the other and the first branching node which
dominates A dominates B.
The present analysis of the ungrammaticality of (16) invites two important questions.
One is what blocks the QM, which is the wh-operator and unselective binder of the indefinite
wh-word, from being raised from I to C in its LF in (27). The other is what blocks
reconstruction of the subject QP into its base position of Spec of VP. If either of the two
options were allowed, QM could eventually take scope over the QP, satisfying (18). The first
option, however, is not a possibility, since the QM in the head of IP in (27) can already bind the
indefinite wh-word for the wh-interrogative reading to obtain, hence discharging the wh-
feature of the QM. Since nothing requires movement of the QM, it should not be raised, given
economy in the sense of Chomsky (1995). The second option is not possible either, since
Korean data to be discussed in section 5.1 involving scrambling strongly suggest that
A-movement is not subject to reconstruction (see Chomsky, 1995; Lasnik, 1999; Mahajan,
1990, among others).21
Now, let us turn to (17) repeated as (28) to see how the lack of the intervention effect can be
accounted for, especially given the proposal that way why is not an indefinite, and that QM is
base-generated in the head of IP.
(28) Amwuto way Mary-lul chotayhaci an hayss-ni?anyone why M-ACC invite NOT did-QM
Why did no one invite Mary?
Note that way, why, not being an indefinite, should move into Spec of CP at LF to be
properly interpreted as a wh-interrogative via spec-head agreement with the QM that should be
raised from its base generated position of I into C at LF to discharge its wh-feature. The LF will
be (29), with ti in the head of IP the trace of the QM and the same ti within VP the trace of way
why, respectively.22
(29) [CP whyi [C QMi [IP QPk [I ti [TP[VP tk V ti ]]]]]]
As one can see above, the wh-interrogative way why in Spec of CP can take scope over the
subject QP in Spec of IP, not violating (18), thus correctly predicting (28) as grammatical.
Now before closing the section, let us turn to (2), where the indefinite wh-word precedes the
NPI without exhibiting the intervention effect. Consider (2a) as a representative one, repeated
below as (30).
Y.-S. Choi / Lingua 117 (2007) 20552076 2067
21 Chomsky (1995:326) claims that A-movement is not subject to reconstruction mainly based on the following (also see
Lasnik, 1999):
(i) *Johni expected himi to seem to me [ti to be intelligent]
Under the interpretation as indicated, (i) can be construed as a Binding Condition B violation, though under
reconstruction the violation should be obviated with him interpreted in the position occupied by the trace (cf. May,
1985; Fox, 2000).22 I agree with an anonymous reviewer in that movement of way why in this paper is not morphology driven. Although
I need to work further on this issue, an alternative proposal one may suggest would be that movement of way why into C
be attracted by the QM in the head of IP via CI. Note that selectional restriction is also satisfied via CI in Korean with
the present proposal for the QM heading IP. If this much is assumed, one may possibly resolve the problem in that
movement of way why is now morphology driven, while QM stays in situ with its [+WH] checked via CI. Although
controversial, the idea does not seem to be that unreasonable, given the status of the QM for way why as a mere
WH-feature holder unlike the QM for indefinite wh-words, which serves as the wh-operator, binding them for proper
interpretation.
(30) Nwu-ka amwuto chotayhaci an hayss-ni?who-NOM anyone invite NOT did-QM
Who invited no one?
One may first wonder how the wh-interrogative reading can obtain given that the indefinite
wh-word is in Spec of IP at the overt syntax and that the QM is base-generated in the head of IP.
Since the indefinite wh-word cannot be reconstructed into the base position of Spec of VP, it
follows that the QM should move into the head of CP at LF as a last resort such that it can bind the
indefinite wh-word for the wh-interrogative reading to obtain, discharging the wh-feature of the
QM. Otherwise the indefinite wh-word cannot be so construed to begin with. As a result, the
following LF for (30) obtains, with the occurrence of ti in the head of IP and the same occurrence
of ti within VP the trace of the QM and that of the wh-word, respectively, and with the object QP
adjoined to VP to avoid a type mismatch23,24:
(31) [CP[C QMi [IP whoi-NOM [I ti [TP[VP QPj -ACC [VP ti V tj ]]]]]]]
As one can see, the QM as the wh-operator binding the indefinite wh-word and marking its
scope can take scope over the object QP in the VP-adjoined position, satisfying the constraint in
(18), and (30) is therefore correctly predicted as grammatical.
5. Prediction of the present analysis
The prediction the present analysis makes is that in wh-questions where the QM as the
wh-operator marking the scope of an indefinite wh-word can take scope over a QP, intervention
effects will somehow vanish even when the QP is followed by the indefinite wh-word. As a matter
Y.-S. Choi / Lingua 117 (2007) 205520762068
23 One may wonder why the grammar does not utilize the spec-head agreement mechanism for the subject wh-word in
Spec of IP in (31) if there are two mechanisms (unselective binding and spec-head agreement) available in Korean. Then
the raising of the QM to C would be superfluous in (31) as an anonymous reviewer points out. Note that wh-words (other
than way why) are indefinites with no inherent wh-feature. Hence, the subject wh-word cannot enter into spec-head
agreement with the QM in the head of IP in (31). Please recall that the indefinitewh-word obtains its quantificational force
by being bound by the QM that serves as its wh-operator.24 One may wonder what if the QM binds the indefinite wh-word at D structure in (31) before the latter is raised into its
surface position of Spec of IP. If so, movement of the QM will be unnecessary, as an anonymous reviewer points out. Lee
(1982), however, reports that the subject and object wh-words in (ia) show subtle asymmetry in that the former but not the
latter can take matrix scope (also see Chung, 1996; Nishigauchi, 1990).
(i) a. Bill-i ne-ekey [CPnwu-ka mwues-ul hayss-nya-ko] mwulepoass-ni?B-NOM you-DAT who-NOM what-ACC did-QM-COMP asked-QM
Who is the person x such that Bill asked you what x did? (Nishigauchi, 1990:35, slightly modified)
b. Bill-un Tom-eykey [CPMary-ka mwues-ul nwukwu-eykey cwuess-nunci] mwuless-ni?B-TOP T-DAT M-NOM what-ACC who-DAT gave-QM asked-QM
What is the thing x such that Bill asked Tom to whom Mary gave x?
If the intuition is correct, the idea of the QM binding the indefinite wh-word at D structure for a wh-interrogative reading
cannot account for this asymmetry, since both indefinite wh-words in (ia) will have embedded scope, bound within VP by
the embedded QM. The reading in (ia) obtains straightforwardly in the present system, with the QM in I of the matrix IP
and the QM in I of the embedded IP respectively binding the subject wh-word in Spec of IP and the object wh-word within
VP at LF (assuming local unselective binding of the indefinite wh-word by the binder, see footnote 5). Incidentally, the
reviewer points out that (ib) admits the matrix scope reading of the direct object wh-word (with a strong accent on it),
although marginal. This further suggests that the idea of the QM binding the indefinite wh-word at D structure is
implausible. One may account for the reading in (ib) under the present system by assuming local scrambling of the subject
and the direct object wh-word to the embedded IP-adjoined position at the overt syntax (see Chung, 1996).
of fact, the prediction seems to be confirmed in wh-questions involving double object
constructions and embedded wh-questions.
5.1. Double object constructions
Consider the following wh-question in (32) involving a double object construction with the
indefinite wh-word nwukwu who preceded by the NPI:
(32) ?John-i amwu-ekey-to nwukwu-lul sokayhaci an hayss-ni?J-NOM anyone-DAT who-ACC introduce NOT did-QM
Who did John introduce to no one?
As one can see, despite the intervening NPI, the intervention effect in (32) does not quite arise,
contrary to Beck and Kims (1997) expectation, but just slightly deviant as compared with (33),
where the NPI and the indefinite wh-word are reversed via scrambling (Chomsky and Lasnik,
1977; Hoji, 1985; Ross, 1967; Saito, 1992, among others).
(33) Nwukwu-lul John-i amwu-ekey-to sokayhaci an hayss-ni?who-ACC J-NOM anyone-DAT introduce NOT did-QM
Who did John introduce to no one?
The lack of the intervention effect in (32) can receive a straightforward account in the present
system, whose LF will be (34), with the indirect object QP (NPI) adjoined to VP to avoid a type
mismatch.
(34) [IP NPj-NOM [I QMi [TP[VP QP-DATk [VP tj V tk whoi-ACC ]]]]]
Please note that as the wh-operator the QM in (34) unselectively binds the indefinite
wh-word and marks its scope, taking scope over the indirect object QP in the VP-adjoined
position, thus not violating the constraint in (18). Hence, (32) is correctly predicted as
grammatical.25
Now, one remaining question is why (33) with the indefinite wh-word scrambled to the front of
the sentence is better than (32), although subtle. If the indefinite wh-word reconstructs into its
base position at LF where the constraint in (18) applies, the grammaticality of (33) should be on a
par with that of (32).
It has been suggested that scrambling in monoclausal constructions as in (33) is not subject to
reconstruction (Beck and Kim, 1997, among others). The claim is in fact supported by (35a) and
(35b) that involve Weak Crossover (Chomsky, 1977; Higginbotham, 1980; Postal, 1971; Wasow,
1972) and Binding Condition C (Chomsky, 1981), respectively26:
Y.-S. Choi / Lingua 117 (2007) 20552076 2069
25 Beck and Kim (1997:343) reports that the example as in (32) is ungrammatical whereas the one as in (33) is perfect. I
agree with them to the extent that (32) is not as good as (33). At the same time, it should be noted that according to native
speakers I consulted, the former is still acceptable, although not perfect.26 The slightly deviant status of (35) is attributed to the fact that ku he does not yield a bound variable reading as easily
as the English counterpart he (see Hoji, 1988; Hong, 1985; Kang, 1988; Suh, 1990, among others).
(35) a. ?Enu kyoswui-lul kui-uy haksayng-i ti chotayhayss-ni?which professor-ACC he-POSS student-NOM invited-QM
*Which professori did hisi student invite?
b. ?[Johni-uy enu sacin-ul]j kui-ka ceil tj sileha-ni?J-POSS which picture-ACC he-NOM most dislike-QM
*Which picture of Johni does hei dislike most?
The grammaticality of (35) strongly indicates that scrambling of a wh-word in a monoclausal
construction is A-movement and that it is not subject to reconstruction.
Now, assuming monoclausal scrambling in (33) involves IP adjunction (see Cho, 1994; Lee,
1993b; Saito, 1992, among others), the QM should move into the head of CP at LF as a last resort
in order to bind the scrambled indefinite wh-word for the wh-interrogative reading to obtain,
discharging the wh-feature of the QM.27 Otherwise, the indefinite wh-word cannot be so
construed to begin with. As a result, the LF for (33) will be (36), with ti in the head of IP and the
same occurrence of ti within VP the trace of the QM and that of the wh-word, respectively.
(36) [C QMi [IP whoi-ACC [IP NPj-NOM [I ti [TP [VP QP-DATk [VP tj V tk ti ]]]]]]]
As one can see, one salient difference between the two LF representations in (34) and (36) is
that a QP intervenes between the QM that serves as the wh-operator and the wh-word that serves
as its restriction in the former but not in the latter. Please recall our discussion in section 3 that an
indefinite wh-word serves as the restriction of the QM, the wh-operator that unselectively binds
it.
On the basis of this difference, I suggest that the slightly deviant status of (32) as compared
with (33) has to do with the fact that QM and its restriction are separated by an intervening
(defined in terms of c-command) QP at LF, albeit the constraint in (18) is satisfied. What I am
suggesting here is reminiscent of Pesetsky (2000:67), where essentially the same idea is
independently proposed as based on English and German data.28 The present suggestion seems to
be plausible, especially given the fact that the example in (37) with way why preceded by the
NPI is perfect. The LF is given in (38), with ti in the head of IP and the same tiwithin VP the trace
of the QM and that of way why, respectively.
(37) John-i amwu-ekey-to way Mary-lul sokayhaci an hayss-ni?J-NOM anyone-DAT why M-ACC introduce NOT did-QM
Why did John introduce Mary to no one?
(38) [CP whyi [C QMi [IP NPk-NOM [I ti [TP [VP QP-DATm [VP tk V tm ti ]]]]]]]
As one can see above in (38), the wh-interrogative way why in Spec of CP takes scope over
the indirect object QP in the VP-adjoined position, satisfying the constraint in (18). Moreover,
since way why does not leave its restriction at LF, the indirect object QP intervening between
way why and its trace does not affect the grammaticality of (37).29
Y.-S. Choi / Lingua 117 (2007) 205520762070
27 It is quite a standard assumption that the IP-adjoined position can serve as an A-position in Korean type languages
(Kuroda, 1988; Saito, 1992; Lee, 1993b, among others). Also see Chomsky (1993:27ff) for related discussions.28 Pesetsky (2000:67) suggests that a semantic restriction on a quantifier (including wh) may not be separated from that
quantifier by a scope-bearing element.29 The same is true of (28) with way why and (30) with an indefinite wh-word in section 4. Both satisfy (18). Also the
wh-word in the former does not leave its restriction but variable at LF, and the QM and its restriction are not intervened by
a QP at LF in the latter.
5.2. Embedded wh-questions
Next, let us turn to the intervention effect in embedded wh-questions, which has not been
discussed fully yet in Korean literature, including Beck and Kim (1997). Interestingly, the
intervention effect in (16) somehow vanishes when embedded as illustrated by (39), which is just
slightly deviant as compared with (40) where the wh-word precedes the NPI.
(39) (?)Ne-nun [CPamwuto nwukwu-lul chotayhaci an hayssta-ko] sayngkakha-ni?you-TOP anyone who-ACC invite NOT did-COMP think-QM
Who do you think no one invited?
(40) Ne-nun [CPnwu-ka amwuto chotayhaci an hayssta-ko] sayngkakha-ni?you-TOP who-NOM anyone invite NOT did-COMP think-QM
Who do you think invited no one?
The lack of the intervention effect in (39) follows in the present system, given its LF in (41):
the QM in the matrix clause, that is, the wh-operator that unselectively binds the indefinite wh-
word in the embedded clause and marks its scope, takes scope over the subject QP in the
embedded Spec of IP, satisfying the constraint in (18), and thus predicting (39) as grammatical.
(41) [IP NPk-TOP [I QMi [TP[VP tk V [CP[IP QPj [TP[VP tj V whoi-ACC ]]]]]]]]
Again the slightly deviant status of (39) is attributed to the fact that the QMand its restriction are
separated by the interveningQP in the embeddedSpec of IP, which is not the case in (40), whoseLF
is in (42). As one can see in (42), the QM in the matrix clause binding the indefinite wh-word in the
embedded clause andmarking its scope takes scope over the object QP in the VP-adjoined position
in the embedded clause, without the QP intervening the QM and its restriction.30
(42) [IPNPk-TOP [I QMi [TP[VP tk V [CP[IP whoi-NOM [TP[VP QPj-ACC [VP ti V tj]]]]]]]]]
The present analysis can also account for the example in (43) with way why, which does not
exhibit the intervention effect. The LF for (43) is given in (44), with ti in the head of IP and the
same ti within VP the trace of the QM and that of way why, respectively:
Y.-S. Choi / Lingua 117 (2007) 20552076 2071
30 According to the present proposal, the following sentences should differ in grammaticality, which an anonymous
reviewer doubts:
(i) a. ?*Amwuto John-eykey [CPMary-ka nwukwu-lul mannassta-ko]anyone J-DAT M-NOM who-ACC met-COMP
malhaci an hayss-ni?
say NOT did-QM
Who did no one tell John that Mary met?
b. ?John-i amwu-eykey-to [CPMary-ka nwukwu-lul mannassta-ko]J-NOM anyone-DAT M-NOM who-ACC met-COMP
malhaci an hayss-ni?
say NOT did-QM
Who did John tell no one that Mary met?
Although subtle, I still find (ib) is somehow better than (ia), with its slight deviant status having to do with the fact that the
QM and its restriction are separated by the intervening QP at LF, although it conforms to the constraint in (18).
(43) Ne-nun [CPamwuto Tom-ul way chotayhaci an hayssta-ko] sayngkakha-ni?you-TOP anyone T-ACC why invite NOT did-COMP think-QM
Why do you think no one invited Tom?
(44) [CPwhyi [C QMi [IPNPk-TOP [I ti [TP[VP tk V [CP[IP QPj [TP[VP tj V ti ]]]]]]]]]]
As one can see in (44), the wh-interrogative way why in Spec of CP in the matrix clause can
take scope over the subject QP in the embedded Spec of IP, not violating the constraint in (18).
Also note that the grammaticality of (43) is better than (39), although subtle. This is because way
why does not leave its restriction at LF, thus the subject QP intervening between the
wh-operator way why and its trace within the VP not affecting the grammaticality of (43).31
By contrast, under Beck and Kims (1997) system, the intricate pattern of intervention effects
in wh-questions with an indefinite wh-word and way why in double object constructions in (32)
and (37), and the embedded wh-question constructions in (39) and (43) remains unexplained.
Under their system assuming LF wh-movement a la Huang (1982), these examples will be ruled
out as equally ungrammatical contrary to the fact, since negation an not which triggers the NIB,
will intervene between the wh-operator in Spec of CP and its variable at LF, violating MNSC as
schematically represented below in (45).
(45) [CP wh-operatori . . .. [VP an [VP . . . ti ]]]
6. Implications for intervention effect with other types of QPs
I thus far claimed that the ungrammaticality of (16) with a subject QP followed by an
indefinite wh-word is essentially due to violation of the LF constraint in (18). Interestingly, the
Y.-S. Choi / Lingua 117 (2007) 205520762072
31 An anonymous reviewer points out that there exists a contrast between the indefinite wh-word and way why in (i)
with the lalter exhibiting the inner island effect (Ross, 1983).
(i) a. John-un [CPMary-ka mwusun kwamok-ul tulessta-ko] sayngkakhaciJ-TOP M-NOM what course-ACC took-COMP think
an ha-ni?
NOT do-QM
What course doesnt John think Mary took?
b. ?*John-un [CPMary-ka swuhak-ul way tulessta-ko] sayngkakhaciJ-TOP M-NOM math-ACC why took-COMP think
an ha-ni?
NOT do-QM
Why doesnt John think Mary took the math the class?
The reviewer thus points out that if negation does not block movement of way why at LF as shown in (28) and (43), (ib)
should be fine, too. However,why in English does not show the inner island effect in (iiab) either, in contrast to (iic) which
shows the effect in that it does not allow a long construal although with a strong accent on dont the effect would disappear
(Christopher Long, personal communication).
(ii) a. Why didnt Mary come to class on time?
b. Why do you think Mary didnt come to class on time?
c. Why dont you think Mary came to class on time?
This suggests that the inner island effect for WHY does NOT show up as in (iiab) whereas the effect does indeed show up
as in (ib) and (iic) whereWHY in the matrix Spec of CP and its trace in the embedded clause are separated by the negation
in the matrix clause at LF. If so, the examples in (iab) further support two ways of scope taking in Korean wh-questions,
with the indefinite wh-word not subject to the inner island effect in contrast to way why since the former does not move
(cf. Kuno and Takami, 1997).
acceptability of (16) improves when the subject QP is replaced with a universal QP as illustrated
in (46).32
(46) ?Motun salam-i nwukwu-lul chotayhayss-ni?everyone-NOM who-ACC invited-QM
Who did everyone invite?
The grammaticality of (46) is quite unexpected, given its LF in (47) where the QM in the head
of IP, which as the wh-operator binds the indefinite wh-word and marks its scope, cannot take
scope over the subject QP in Spec of IP, in violation of the constraint in (18).
(47) [IPQPj-NOM [I QMi [TP [VP tj V whoi-ACC ]]]]
One needs to account for why (46), although violating (18), is still acceptable in contrast to
(16). The grammaticality of (46), as I claim, is due to nonquantificational specific group
interpretation of the universal QP, thus circumventing (18) in contrast to (16) with the existential
QP amwuto no one, which resists the construal.33
Now the prediction our analysis makes is that when the QP in (46) is replaced with a QP such
as manun NP many NP, se NP three NP, myess myess NP several NP or taypwupwun-uy NP
most of NP that can have a specific group construal, the grammaticality will remain more or less
the same, whereas with a modified QP such as se NP miman less than three NP or twu NP isang
more than two NP which are well known for resisting a specific group construal, (46) will
become ungrammatical (Beghelli, 1997; Liu, 1997; Reinhart, 1997, among others).34 The
prediction is confirmed as shown in (48) and (49).35,36
(48) a. ?Manun salam-i nwukwu-lul chotayhayss-ni?many man-NOM who-ACC invited-QM
Who did many men invite?
Y.-S. Choi / Lingua 117 (2007) 20552076 2073
32 See Kim (1991) for the same intuition on the example as in (46) (also see Hoji, 1985, for Japanese).33 Although the term specific tends to be applied typically in relation to indefinites, it can also apply to definites (see
Lyons, 1999:168ff).34 Although the semantics of isang the same or larger in number or amount is different from more than in English, I
will simply translate it as more than for simplicitys sake.35 most (of ) NP type QP also seems to admit a group construal as shown in (i).
(i) a. John didnt like most students.
b. Many people didnt criticize most of the presenters.
The example in (ia) means that there is a specific group of most students such that John did not like them. Also the QP
most of the presenters in (ib) does not covary with the choice of the individuals of the set denoted by the QP many people,
while taking scope over negation (see Sohn, 1995:166168).36 An anonymous reviewer notes that the following with a modified subject QP seems fine:
(i) Ipen taysen-eyse kwapanswu-ka nwukwu-lul cicihayss-ni?this presidential election-in more than half-NOM who-ACC supported-QM
Who did more than half of the voters support in the presidential election?
I suggest that its improved grammaticality has to do with the discourse-linking effect as discussed in footnote 2. Please
note that the presidential election is a situation where the indefinite wh-word is most likely to be discourse-linked. As the
reviewer points out, given that an indefinite wh-word does not move even at LF in the present system, one cannot however
represent the licensing mechanism of a discourse-linked wh-word as distinguished from a non-discourse linked wh-word
in wh-questions as in Pesetsky (1987). The intuition, however, seems to be fairly clear: unlike a non-discourse linked
wh-word, a discourse-linked wh-word, being nonquantificational even under the wh-interrogative reading, may not be
subject to the constraint in (18).
b. ?Taypwupwun-uy salam-i nwukwu-lul chotayhayss-ni?most-POSS man-NOM who-ACC invited-QM
Who did most of the men invite?
(49) a. *Twu salam isang-i nwukwu-lul chotayhayss-ni?two man more-NOM who-ACC invited-QM
Who did more than two men invite?
b. *Se salam miman-i nwukwu-lul chotayhayss-ni?three man less-NOM who-ACC invited-QM
Who did less than three men invite?
If our analysis is one the right track, (49) will also improve in grammaticality when
embedded.37 The prediction is indeed confirmed as illustrated in (50), whose LF is given in (51).
(50) (?)Ne-nun [CPtwu salam isang-i nwukwu-lul chotayhayssta-ko] sayngkakha-ni?you-TOP two man more-NOM who-ACC invited-COMP think-QM
Who do you think that more than two men invited?
(51) [IPNPk-TOP [I QMi [TP[VP tk V [CP[IP QPj-NOM [TP[VP tj V whoi-ACC ]]]]]]]]
Again, the slightly deviant status of (50) is attributed to the fact that QM as the wh-operator in
the matrix clause and the indefinite wh-word in the embedded clause that serves as the restriction
of the wh-operator are separated by the intervening QP in the embedded clause at LF, although it
satisfies the constraint in (18).
I would like to close the section by noting that the contrast in grammaticality between (49) and
(50) again poses a nontrivial problem to Beck and Kim (1997). It is not clear how they can account
for the contrast in grammaticality. It will not help to stipulate that modified QPs in (49) are
interveners for wh-movement at LF as in Beck (1996), since (50) with the same type of QP is still
acceptable.
7. Conclusion
The intervention effect in Korean wh-questions can receive a natural account, given the
proposed twoways of scope taking of in situwh-words: in situ interpretation of indefinitewh-words
via unselective binding by the question morpheme versus movement of way why. Based on the
interesting contrast of the intervention effect in wh-questions with the two types of wh-words in
various constructions, for which Beck and Kim (1997) cannot offer a principled account, the
present research claims that the intervention effect is not a movement but scope phenomenon, as
opposed to them. The present research, if on the right track,may lead us to a better understanding of
the nature of the intervention effect in other wh-in-situ languages typologically akin to Korean.
Acknowledgements
Among others, I would like to express my special thanks to the anonymous reviewers of
Lingua for all their valuable feedbacks on this paper. One can certainly see their influence
Y.-S. Choi / Lingua 117 (2007) 205520762074
37 An anonymous reviewer reports that the grammaticality does not improve in (49) even when the indefinite wh-word is
replaced with way why, contrary to the prediction of the present proposal. My intuition rather says the grammaticality
improves.
everywhere in this paper. I also would like to thank Nam-kil Kim, Chungmin Lee, Christopher
Long, Barry Schein, Jean-Roger Vergnaud and Maria Luisa Zubizarreta for their comments and
encouragement at various stages of this paper.
References
Aoun, J., Li, A., 1993. Wh-elements in situ: syntax or LF? Linguistic Inquiry 24, 199238.
Baker, L., 1970. Notes on the description of English questions: the role of an abstract question morpheme. Foundations of
Language 6, 197219.
Beck, S., 1996. Quantified structures as barriers for LF movement. Natural Language Semantics 4, 156.
Beck, S., Kim, S.-S., 1997. Wh- and operator scope in Korean. Journal of East Asian Linguistics 6, 339384.
Beghelli, F., 1997. The syntax of distributivity and pair-list readings. In: Szabolcsi, A. (Ed.), Ways of Scope Taking.
Kluwer Academic Publishers, Dordrecht, pp. 349408.
Cheng, L., L-S., 1991. On the typology of wh-questions. Doctoral Dissertation. MIT, Cambridge, MA.
Chierchia, G., 1991. Functional wh and weak crossover. In: Bates, D. (Ed.), Proceedings of the 10th West Coast
Conference on Formal Linguistics. CSLI, Stanford, pp. 7590.
Chierchia, G., 1993. Questions with quantifiers. Natural Language Semantics 1, 181234.
Cho, E., 1998. Why, contrastive topic and LF movement. In: Silva, D. (Ed.), Proceedings of the Japanese Korean
Linguistics, vol. 8. CSLI, Stanford, pp. 403415.
Cho, J.-H., 1994. On scrambling: reconstruction, crossover and anaphor binding. In: Kim-Renaud, Y.-K. (Ed.),Theoretical
Issues in Korean Linguistics. CSLI, Stanford, pp. 255273.
Choe, H.-S., 1995. Focus and topic movement in Korean and licensing. In: Kiss, K.E. (Ed.), Discourse Configurational
Languages. Oxford University Press, New York, pp. 269334.
Choi, H.-P., 1937. Wuli Malpon (Korean Grammar). Yenhuycenmwun Press, Seoul.
Choi, Y.-S., 2002. Asymmetry of scope taking in wh-questions. Doctoral Dissertation. University of Southern California,
Los Angeles.
Chomsky, N., 1955. The Logical Structure of Linguistic Theory. (Plenum Press, New York, 1975).
Chomsky, N., 1973. Conditions on transformations. In: Anderson, S., Kiparsky, P. (Eds.), A Festschrift for Morris Halle.
Holt, Rinehart and Winston, New York, pp. 232286.
Chomsky, N., 1977. Conditions on rules of grammar. In: Essays on Form and Interpretation, Elsevier North-Holland, New
York, pp. 163210.
Chomsky, N., 1981. Lectures on Government and Binding. Foris, Dordrecht.
Chomsky, N., 1993. A minimalist program for linguistic theory. In: Hale, K., Keyser, S. (Eds.), The View from Building
20: Essays in Linguistics in Honor of Sylvain Bromberger. MIT Press, Cambridge, MA, pp. 152.
Chomsky, N., 1995. The Minimalist Program. MIT Press, Cambridge, MA.
Chomsky, N., Lasnik, H., 1977. Filters and control. Linguistic Inquiry 8, 425504.
Chung, D., 1996. On the representation and licensing of Q and Q-dependents. Doctoral Dissertation. University of
Southern California, Los Angeles.
Chung, T., 1991. On wh-adjunct extraction in Korean. In: Kuno, S., et al. (Eds.), Harvard Studies in Korean Linguistics,
vol. 4. Hanshin Publishing Company, pp. 241248.
Engdahl, E., 1986. Constituent Questions. D. Reidel Publishing Company, Dordrecht/Boston/Lancaster/Tokyo.
Fox, D., 2000. Economy and Semantic Interpretation. MIT Press, Cambridge, MA.
Grimshaw, J., 1986. Some Consequences of the IP/CP Hypothesis. Brandeis University, MS.
Heim, I., 1982. The semantics of definite and indefinite noun phrases. Doctoral Dissertation. University of Massachusetts,
Amherst.
Heim, I., Kratzer, A., 1998. Semantics in Generative Grammar. Blackwell Publishers, Malden, MA.
Higginbotham, J., 1980. Pronouns and bound variables. Linguistic Inquiry 11, 679708.
Hoji, H., 1985. Logical form constraints and configurational structures in Japanese. Doctoral Dissertation. University of
Washington, Seattle.
Hoji, H., 1988. On the so-called overt pronouns in Japanese and Korean. In: Baek, E.-J. (Ed.), Papers From the Seventh
International Conference on Korean Linguistics, pp. 6177.
Hong, S.-S., 1985. A and A binding in Korean and English: government and binding parameters. Doctoral Dissertation.
University of Connecticut, Storrs.
Huang, C.-T.J., 1982. Logical relations in Chinese and the theory of grammar. Doctoral Dissertation. MIT, Cambridge,
MA.
Y.-S. Choi / Lingua 117 (2007) 20552076 2075
Joo, S.Y., 1989. A cross-linguistic approach to quantification in syntax. Doctoral Dissertation. University of Wisconsin-
Madison.
Kang, M.-Y., 1988. Topics in Korean syntax: phrase structure, variable binding and movement. Doctoral Dissertation.
MIT, Cambridge, MA.
Karttunen, L., 1977. Syntax and semantics of questions. Linguistics and Philosophy 1, 344.
Katz, J., Postal, P., 1964. An Integrated Theory of Linguistic Descriptions. MIT Press, Cambridge, MA.
Kim, S., 1991. Chain scope and quantification structure. Doctoral Dissertation. Brandeis University.
King, R., 1994. History of reported speech in Korean. Korean Linguistics 8, 138.
Koopman, H., Sportiche, D., 1991. The position of subjects. Lingua 85, 211258.
Kuno, S., Takami, K., 1997. Remarks on negative islands. Linguistic Inquiry 28, 553576.
Kuroda, S.-Y., 1965. Generative grammatical studies in the Japanese language. Doctoral Dissertation. MIT, Cambridge,
MA.
Kuroda, S.-Y., 1988.Whether we agree or not: a comparative syntax of English and Japanese. Lingvisticae Investigationes
12, 147.
Ladusaw, W., 1980. Polarity Sensitivity as Inherent Scope Relations. Garland Press, New York.
Lasnik, H., 1999. Chains of arguments. In: Epstein, S., Hornstein, N. (Eds.), Working Minimalism. MIT Press,
Cambridge, MA, pp. 189215.
Lee, H.-S., 1982. Asymmetry in Island Constraints in Korean. UCLA, MS.
Lee, P.-Y., 1993a. Kwukeuy Inyongkwumwun Yeonkwu (A Study on Reported Speech in Korean). Tower Press, Seoul.
Lee, Y.-S., 1993b. Scrambling as case-driven obligatory movement. Doctoral Dissertation. University of Pennsylvania.
Lewis, D., 1975. Adverbs of quantification. In: Keenan, E. (Ed.), Formal Semantics of Natural Language. Cambridge
University Press, Cambridge and New York, pp. 315.
Li, A., 1992. Indefinite wh in Mandarin Chinese. Journal of East Asian Linguistics 1, 125155.
Liu, F.-H., 1997. Scope and Specificity. John Benjamins Publishing Company, Amsterdam and Philadelphia.
Lyons, C., 1999. Definiteness. Cambridge University Press, Cambridge.
Mahajan, A., 1990. The A/A distinction and movement theory. Doctoral Dissertation. MIT, Cambridge, MA.
Martin, S., 1992. A Reference Grammar of Korean: a Complete Guide to the Grammar and History of the Korean
Language. Charles E. Tuttle Publishing Company, Rutland/Vermont/Tokyo, Japan.
May, R., 1985. Logical Form. MIT Press, Cambridge, MA.
May, R., 1988. Ambiguities of quantification and wh: a reply to Williams. Linguistic Inquiry 19, 118135.
Nishigauchi, T., 1990. Quantification in the Theory of Grammar. Kluwer Academic Publishers, Dordrecht.
Pesetsky, D., 1987. Wh in situ: movement and unselective binding. In: Reuland, E., ter Meulen, A. (Eds.), The
representation of (In)definiteness. MIT Press, Cambridge, MA, pp. 98129.
Pesetsky, D., 2000. Phrasal Movement and its Kin. MIT Press, Cambridge, MA.
Pollock, J.-Y., 1989. Verb movement, universal grammar, and the structure of IP. Linguistic Inquiry 20, 365424.
Postal, P., 1971. Cross-over Phenomena. Holt, Reinhart and Winston, New York.
Reinhart, T., 1976. The syntactic domain of anaphora. Doctoral Dissertation. MIT, Cambridge, MA.
Reinhart, T., 1997. Quantifier scope: how labor is divided between QR and choice functions. Linguistics and Philosophy
20, 335397.
Ross, J., 1967. Constraints on variables in syntax. Doctoral Dissertation. MIT, Cambridge, MA.
Ross, J., 1983. Inner Islands. MIT, MS.
Saito, M., 1992. Long distance scrambling in Japanese. Journal of East Asian Linguistics 1, 69118.
Sells, P., 2001. Negative polarity licensing and interpretation. In: Kuno, S., et al. (Eds.), Harvard Studies in Korean
Linguistics, vol. 9. Hanshin Publishing Company, pp. 322.
Sohn, H.-M., 1999. The Korean Language. Cambridge University Press, Cambridge.
Sohn, K.-W., 1995. Negative polarity items, scope and economy. Doctoral Dissertation. University of Connecticut, Storrs.
Song, S.-O., 1996. On the development of sentence-final particles in Korean. In: Akatsuka, N., et al. (Eds.), Proceedings
of the Japanese Korean Linguistics, vol. 5. CSLI, Stanford, pp. 219234.
Suh, J.-H., 1990. Scope phenomena and aspects of Korean syntax. Doctoral Dissertation. University of Southern
California, Los Angeles.
Suh, J.-M., 1987. A Study on the Interrogative Sentences in Korean. Tower Press, Seoul.
Tsai, W.-T.D., 1994. On economizing the theory of A-bar dependencies. Doctoral Dissertation. MIT, Cambridge, MA.
Wasow, T., 1972. Anaphoric relations in English. Doctoral Dissertation. MIT, Cambridge, MA.
Whitman, J., 1989. Topic, modality, and IP structure. In: Kuno, S., et al. (Eds.), Harvard Studies in Korean Linguistics,
vol. 3. Hanshin Publishing Company, pp. 341356.
Y.-S. Choi / Lingua 117 (2007) 205520762076