177

Overt Case and Covert Case in Korean*46

Yong-Ha Kim
(Keimyung University)

In this paper, I concern myself with the relationship of overt and

covert Case to two types of word order in Korean; free vs. rigid word
order. To characterize this relationship I will essentially assume the
Attract-F theory in Chomsky 1995. I assume that (overt) Case markers
are Ds with phonological features and Case features, and phono-
logically null (covert) Case markers also exist as their counterparts.
These two types of Case markers are heads of DPs and hence subject
to feature movement. Assuming that the overt Case markers indicate
the strong D-feature of I and v while the covert Case markers signify
the lack of the strong D-feature of I and v, we can successfully
explain the contrast between overt Case and covert Case in Korean.

1. Introduction

In Korean, grammatical functions of NPs are manifested by Case markers.

*This paper was partially presented at the monthly workshop of the Korean Generative
Grammar Circle on 22 November 1996. I am very grateful to all the participants, especially
to Dr. Dong-Whee Yang, Dr. Keun-Won Sohn, Prof. Sun-Woong Kim, Prof. Yong-Suck
Kim, and Prof. Sook-hee Lee for their helpful comments. I would also like to particularly
acknowledge Prof. Hee-Don Ahn’s deep interest in this paper and his encouragement and
interesting suggestions. I have benefited from two anonymous reviewers. Their sharp criticism
and helpful suggestions made this paper much improved. Needless to say, all errors are mine.
178 Yong-Ha Kim

Due to the Case markers, Korean shows free word order, which in turn
leads some grammarians to classify it as a non-configurational language.
Since Saito 1985, however, it has been claimed that free word order
languages such as Korean and Japanese have the same configurational
structure as English, and this claim has been established in generative
grammar. According to this claim, free word order in Korean (or Japanese)
has to be treated as a result of the syntactic movement scrambling.
There is additional evidence supporting the configurationality of Korean.
When argument NPs appear without overt Case markers in a sentence, the
relative word order is very rigid. In other words, Korean shows two types
of word order.

(1) a. Chelswu-ka ku chayk-ul ilk-nun-ta.

-Nomthe book-Acc read
‘Chelswu reads the book.’
b. Ku chayk-ul Chelswu-ka ilk-nun-ta.
the book Chelswu read
(2) a. Chelswu ku chayk ilk-nun-ta.1
*
b. Ku chayk Chelswu ilk-nun-ta.

If free word order is crucial evidence for non-configurationality, the fact

that Korean also has the rigid word order seems to be counterevidence.
This paper, however, does not aim to discuss the configurationality of
Korean. I do not doubt the configurationality of Korean, but I concern
myself with the two types of Korean word order (free vs. rigid) and with

1. An anonymous reviewer points out that the grammaticality of (2a) is doubtful. He

seems to regard the sentence (2a) as ungrammatical or as marginal at best. Even though
(2a) seems to have a different status from (1a), most informants I have consulted judged
it grammatical.
 Overt Case and Covert Case in Korean 179

their relationship to Case markers. The main questions to be answered in

this paper are the following, with respect to the sentences in (1)-(2).

(3) a. Why is word order rigid in Korean when every argument NP has
no Case marker?
b. Why is word order free in Korean when every argument NP has
a Case marker?
c. Why cannot free word order appear even in case that only one of
the argument NPs has no Case marker?

To answer these questions, I will take a minimalist approach, which will

uncover the mystery of the Korean Case phenomena. Section 2 surveys
Chomsky’s (1995) Minimalist Program. Section 3 presents the relevant data
in Korean, and suggests that Korean has two kinds of Case markers which
belong to the category D―overt and covert―and of these two kinds of
Case markers, only the overt Case marker is subject to movement in overt
syntax. In section 4, conclusions are presented.

2. The Minimalist Program

Chomsky (1992) presents a new model of grammatical theory in which

he explicitly refines the least effort and last resort principles in Chomsky
1991. This new model requires that every grammatical operation be based
on conceptual necessity, and that every grammatical operation create
representations that contain only legitimate elements at the interface levels.
Based on conceptual necessity, Chomsky eliminates D- and S-Structure
entirely, both of which do not have any independent significance. The only
180 Yong-Ha Kim

necessary levels are interface levels, PF (Phonetic Form) and LF (Logical

Form), which in turn are related to two performance systems, the
articulatory-perceptual (A-P) system and the conceptual-intentional (C-I)
system, respectively.
Chomsky (1992) also claims that every syntactic derivation is determined
by the strength of the morphological features of functional elements. Thus,
syntactic derivations take place if and only if some morphological features
have to be checked off for convergence at the interface levels. Such
morphological features are Case features, agreement features (or ø-features),
tense features, and so on. These features are either strong or weak and the
same feature may have different strength from language to language (and
from element to element). If strong morphological features are not checked
off before Spell-Out2, the derivation crashes because unchecked strong
features are not legitimate at PF. On the other hand, if weak morphological
features are checked off before Spell-Out, the derivation is blocked by the
convergent derivation with weak morphological features unchecked up to
Spell-Out. In this case, the factor that chooses a derivation over another is
not the interface conditions but the economy consideration. The quicker a
derivation’s access to LF (or PF), the more economical the derivation. See
the following famous examples from Pollock 1989:

(4) a. John often kisses Mary.

b. *Jean souvent embrasse Marie.

*
(5) a. John kisses often Mary.
b. Jean embrasse souvent Marie.

2. Spell-Out is the branching point into PF and LF. It corresponds to the S-structure
in the GB model of syntax.
 Overt Case and Covert Case in Korean 181

Whereas Pollock (1989) explains the difference between English and

French verb positions by assuming that English Agr is Ɵ-opaque but French
Agr is Ɵ-transparent, Chomsky (1992) posits the strong V-feature for Agr in
French and the weak V-feature for Agr in English. As evidenced by the
relative order of the verbs and adverbs in (4) and (5), the English verb kiss
must not move at the overt syntax because of Agr’s weak V-feature, but
the French verb embrasse must move up to Agr at the overt syntax because
of Agr’s strong V-feature. Notice that whereas the ungrammaticality of (4b)
is due to the unchecked strong feature of embrasse, and hence PF-crash, the
ungrammaticality of (5a) isn’t due to PF-crash. In fact, both (4a) and (5a)
are produced through convergent derivations. What blocks (5a) is (4a), the
more economical derivation than (5a)3.
There is another difference between Pollock 1989 and Chomsky 1992. In
Pollock’s approach, inflectional categories are actual affixes. Hence, the
inflectional affixes lower to the verb in English, but remain in situ in
French to host the verb. But Chomsky departs from this approach, assuming

3. There are many other conditions on economy of derivation some of which are
listed below (cf. Yang 1994).
) Minimal Link Condition (MLC)
Minimize chain links.
a. Argument movement across argument positions is impossible.
b. Operator movement across non-argument positions is impossible.
c. Head movement across head positions is impossible.
ii) Minimal Chain condition
Minimize the length of a chain.
iii) Principle of Fewest Steps
Minimize steps of a derivation.
iv) Greed
An element can move iff the movement fulfills its need.
v) Procrastinate
Apply rules as late as possible.
182 Yong-Ha Kim

that lowering is not permissible at all4. He, instead, suggests that every
inflectional affix is attached to a host at the numeration5. This means that
syntactic movement is only for checking and morphological merger is not
the result of syntactic movement6.
This suggestion directly carries over to DP(or NP)7 movement. All
features that belong to D should be attached to D at the numeration, and
DP movement takes place in order to have the features checked off. D’s
relevant features to be checked are Case features and ø-features. A subject
DP, for example, moves to the Spec of some functional category to satisfy
the Extended Projection Principle.8 In Chomsky 1992, it is argued that the
functional category which provides the subject DP with the Spec position is
Agr. Furthermore, there is another Spec position needed for the object DP.

4. Chomsky (1957) originally suggests that inflectional affixes are syntactic elements
and that these affixes hop to (i.e. lower to) verbal elements. For this, Chomsky has
usually been classified into weak Lexicalists, as opposed to strong Lexicalists who claim
that inflectional morphology, as well as derivational morphology, belongs to the Lexicon.
5. The term numeration is introduced in Chomsky 1994. It forms the reference set,
which is the set of possible derivations.
6. Baker’s (1988) Mirror Principle may still be effective in the minimalist program. In
fact, the hierarchical order of functional categories is determined by the linear order of
affixes that are attached to the verb (See Chomsky 1992, Halle & Marantz 1993). But
there can be some difficulty with the Mirror Principle in Korean syntax. If every
inflectional morphology mirrors syntactic derivations, Korean, with its very rich verbal
morphology, has more articulated structure than English. Moreover, it is not a trivial
matter to determine which affix belongs to which inflectional category. Because this
problem is beyond the scope of this paper, I will not discuss it here.
7. I accept the DP-Hypothesis from Fukui & Speas 1986. In the following, I will use
DP and NP somewhat freely to denote nominal constructions. But, if it is necessary to
draw the distinction between them, I will identify each of them.
8. According to Chomsky (1992), EPP is just a descriptive term. It is I’s strong
D-feature that makes the subject DP move. The reason that the subject DP moves to
[SPEC, AgrS] is that I doesn’t have any available Spec (but see Jonas and Bobaljik
1993).
 Overt Case and Covert Case in Korean 183

This Spec position is also provided by another Agr. Chomsky (1992) calls
these functional elements AgrS and AgrO, respectively.9 But according to
Chomsky (1995), Agr has a very suspicious status. It is not certain that it
has any significant role in the grammar. For example, though it provides
the Spec position for the subject DP to satisfy EPP, the real things that
check off the features of the subject DP are I for EPP (and the nominative
Case) and V for ø-features. Agr is just a kind of place holder. This fact
leads Chomsky (1995) to eliminate the category of Agr.
Chomsky (1995) makes a further step toward the minimalist objective. He
claims that every (syntactic) movement is essentially feature movement and
the movement of a full category is determined by the PF requirements of
individual languages. According to him, the following distinctions among
features are operative at computation.

(6) a. categorial features

b. ø-features
c. Case features
d. strong F, where F is categorial.

He further claims that Agr exists only when it has strong features and
that it is nothing more than an indication of a position that must be
occupied at once by overt operations. This claim actually means that Agr
doesn’t have to exist. Hence, in Chomsky’s (1995) system EPP is satisfied
by the movement of the subject DP to the Spec of I,10 and (the formal
features of) the object DP moves to (the Spec of) the light verb v which

9. In fact, AgrS and AgrO are not different elements. They are two realizations of the
same element, Agr. It is just for descriptive purpose that they are distinguished.
10. I(nfl) is used here instead of T in Chomsky 1995 because there are disputes on
which material checks the nominative Case in Korean. See Yang 1996.
184 Yong-Ha Kim

also provides the configuration for the assignment of the external theta-role.
The basic sentence structure (at least for the transitive construction) in
Chomsky 1995 is as follows.

(7) IP

Spec I'
Subj
I vP

Spec v'
Obj
tSubj v'

v VP

V tObj

In this structure, I attracts Subj to satisfy EPP, and v attracts Obj for the
(overt) checking of some (strong) features. Basically, in Chomsky 1995,
movement takes place only when it is to satisfy the attractor’s need. For
this reason, this theoretical framework is called the Attract-F theory. In the
Attract-F theory, formal features of lexical items are classified with respect
to their Interpretability. A [+Interpretable] feature enters into interpretation at
LF, and hence, is always visible at LF. A [-Interpretable] feature does not
enter into interpretation at LF, and hence, must be eliminated to satisfy the
Full Interpretation Principle. Feature checking is the operation eliminating
[-Interpretable] features. Among the features specified in (6), the strong
categorial features determine the overt movement.11 If a lexical item LI

11. We will see in section 3.3. what the nature of the strong feature is.
 Overt Case and Covert Case in Korean 185

enters the numeration with a strong feature, then it must be eliminated

overtly by checking. If LI has a [-Interpretable] feature but not a strong
feature, then it should be eliminated by the checking procedure, covertly in
this case. [+Interpretable] features and [-Interpretable] features are as follows
(cf. Yang 1995).

(8) [+interpretable]: Formal Features that enter into interpretation at LF.

a. Categorial features
b. ø-features of N
c. Q-features of C

(9) [-interpretable]: Formal Features that do not enter into interpretation at

LF.
a. Case features
b. ø-features of V, A, etc
c. strong feature
d. affix feature

In this paper, I will essentially take the Attract-F theory, but there are
some differences between Chomsky 1995 and this paper. They will be
presented as the discussion proceeds.

3. Overt Case and Covert Case in Korean

Ahn (1966) observes that Korean has sentences containing NP arguments

without Case markers, and argues that they are Case-marked by syntagmatic
relations. He finally proposes that in Korean, there is a mechanism for
186 Yong-Ha Kim

Case-marking which he calls Casus Indefinitus (indefinite Case). Casus Indefinitus

is different from Casus Definitus marked by Case-markers and is a
completely independent Case system in Korean. See the following examples.

(10) a. Enni ku chayk ilk-nun-ta.

(My) sister the book read-Pres-M
‘My sister reads the book’
b. Enni-ka ku chayk-ul ilk-nun-ta.
-Nom -Acc

(11) a. Enni ku chayk-ul ilk-nun-ta.

b. Enni-ka ku chayk ilk-nun-ta.

(12) a. *Ku chayk enni ilk-nun-ta.

b. Ku chayk-ul enni-ka ilk-nun-ta.
*
c. Ku chayk-ul enni ilk-nun-ta.
d. *Ku chayk Enni-ka ilk-nun-ta.

Now, let’s take Ahn’s proposal with a different terminology. I will refer
to Ahn’s Casus Indefinitus as covert Case, and Casus Definitus as overt
Case12. As Ahn (1966) points out, in (10a) and (12a) in which the two
arguments of the verb ilk- occur with the covert Case, the relative order of
them is severely restricted. By contrast, (10b) and (12b), in which both of
the two arguments of the verb ilk- occur with the overt Case, show free
word order. It is noticeable that, when either the subject or the object
occurs with the covert Case, the word order is frozen as shown in (11) and

12. There have been a few Korean linguists interested in overt Case and covert Case.
For detailed discussion of the covert Case in Korean, see Yi 1988 and Y-H Kim 1991
among others.
 Overt Case and Covert Case in Korean 187

(12c-d). Why is it that not only when every argument occurs with the
covert Case, but also when either of them occurs with the covert Case, are
the word orders of the sentences similarly restricted?
In order to answer this question, we should determine the syntactic
positions of the NPs with the covert Case and the NPs with the overt Case.

3.1. Topic, IP-subject, and vP-subject

As mentioned in note 1, an anonymous reviewer doubts the

grammaticality of (10a) (and probably that of (11a)). According to his
judgement, (10a) and (11a) are bad because the nominative Case marker
cannot be dropped whereas the accusative Case marker can. Furthermore, he
claims that his judgement is generally accepted in the discipline of
grammar. In fact, there have been a few generative grammarians who
observe the nominative-accusative asymmetry with respect to Case drop. For
example, Saito (1983) claims that whereas the accusative Case marker -o
can be dropped, the nominative Case marker -ga cannot be dropped in
Japanese because the nominative is an inherent Case which is not assigned
by any element but the accusative is an abstract Case assigned by transitive
verbs. A subject NP, thus cannot go through the Case Filter if the
nominative Case marker is dropped, whereas an object NP can do so since
it is assigned an abstract Case by a Case assigner.
Saito (1983) also presents an example of the nominative Case drop in
Japanese.

(13) John kita no?

came Q
‘Did John come?’
188 Yong-Ha Kim

Following Kuno (1973), Saito explains this seeming exception by arguing

that John is not the subject but the Topic of the sentence. This is
evidenced by the following examples.

(14) a. Dare-ga kita-no?

who-Nom came-Q
‘Who came?’
b. *Dare-wa kita-no?
-Top
*
c. Dare kita-no?

According to Saito, wh-phrases in Japanese can appear as the subject but

not as the Topic, thus (14c) is ungrammatical because the wh-phrase
occupies the Topic position.
However, Saito’s (1983) argument seems not applicable to Korean. The
informants I have consulted accept the sentences containing a subject
without the nominative Case marker as grammatical and such Korean
generative grammarians as Han (1987), Kim (1990, 1991), Ahn (1988), Yim
(1983, 1985) report that the nominative Case drop in Korean is all right
and even prevalent in colloquial speech13. For example, Han (1987) presents

13. The scholars show different views about the effect and mechanism of Case drop.
Han (1987) suggests that the possibility of Case drop is closely related to the speech
level, so the more informal the speech level is, the easier it is to drop Case markers.
Kim (1990) similarly treats the phenomenon of Case drop as outside the domain of
formal grammar. Their viewpoint, however, is very problematic with respect to the
contrasts the data in (10)-(12) show. Contrastively, Yim (1983) claims that Korean Case
markers including both nominative and accusative are dummy affixal elements. According
to him, sentences containing NP arguments without Case markers are more basic and
should be treated as reflecting the configurationality of Korean. Although I do not fully
accept his claim since it is made within the framework of the so-called GB theory, it is
 Overt Case and Covert Case in Korean 189

an apparent counterexample to this argument. Unlike Japanese, Korean

seems to allow the Case drop from a wh-phrase in the subject position.

(15) a. Mwue philyoha-ni?

what necessary-Q
‘What is necessary?’
b. Mwue-ka philyoha-ni?
-Nom
*
c. Mwue-nun philyoha-ni?
-Top

Surprisingly enough, Korean is quite different than Japanese with respect to

the relationship of wh-phrases to nominative and Topic markers. From the
contrasts in these sentences, we can claim that Saito’s (1983) argument
cannot explain the aspects of Case drop in Korean.
Ahn (1996) points out that Han’s (1987) examples all involve
unaccusative verbs and the subject in (15) is a quasi-complement directly
selected by the verb. Thus, he assumes that the contrasts in (15) are in
accordance with Hong’s (1994) Generalization.

(16) Hong’s Generalization

a. Case markers may be dropped in the complement position.
b. Case markers may not be dropped in the specifier position.

Notice that Hong’s Generalization does not depend upon the classification of
Case as Saito’s (1983) argument. According to this generalization, the
possibility of Case drop is determined structurally. Thus, the contrasts in

very similar to the claim made in this paper.

190 Yong-Ha Kim

(15) can be explained by Hong’s Generalization if we admit verbs like

philyoha- are unaccusatives. Furthermore Hong’s Generalization seems to
explain the following data.14

(17) a. *Nwukwu chayk-ul sa-ass-ni?

Who book buy-Past-Q
‘Who bought a book?’
*?
b. Chelswu-ka nwukwu yeyppu-ta-ko sayngkakha-ni?
pretty-M-Comp think
‘Who does Chelswu think is pretty?’

Nwukwu in (17a) should be in a specifier position because it is the subject

of the sentence, and the nominative Case marker cannot be dropped in this
position. Hong’s Generalization seems to correctly predict the deviance of
(17a). (17b) is an ECM construction, so it is likely that nwukwu in this
sentence is in a specifier position so that it is ruled out by Hong’s
Generalization.

14. Hong’s (1994) original examples involve an NP construction which is exemplified

below.

i) *Cekkwun tosi-uy phakoi

enemy city-Gen destruction
‘The enemy’s destruction of the city.’

Though I will not discuss the covert genitive Case, it should be noted that I think i)
has a different status from the data dealt with in this paper because of the peculiarity of
nouns’ theta-role assignment to its arguments. If Grimshaw’s (1990) argument is correct
that nouns cannot assign the external theta-role, cekkwun in i) is not an argument of
phakoi. There are some conditions on the occurrence of the covert Case, of which it is
one that only arguments of a predicate can appear with the covert Case. See Y-H Kim
1991 for some helpful discussions.
 Overt Case and Covert Case in Korean 191

Hong (1994) explains his generalization by incorporation. To recapitulate

his explanation, complement-to-head incorporation is all right because it does
not violate any principle in grammar, but specifier-to-head incorporation is
not permissible because it violates important grammatical principles like
ECP. Though I will not accept his explanation for several reasons,15 it is
desirable to adapt the insight of his generalization to the minimalist
program.
It is permissible to assume that DPs with overt Case are in some
VP-external checking positions, whereas it is not possible to posit a
VP-external position for DPs with covert Case because Case markers cannot
be dropped in this position in Hong’s (1994) framework. In the minimalist
program, for a DP to have its Case feature checked, it must move into the
checking domain of some functional category (whatever the functional

15. First of all, his assumption is too strong that every Caseless N incorporates to a
head governing it. As shown below, certain materials can intervene between a Caseless
NP and a verb (cf. Kim 1991: 20).

i) a. Chelswu-ka hakkyo ohwu-ey/pes-lo ka-n-ta.

-Nom school afternoon/bus-by go-Pres-M
‘Chelswu goes to school in the afternoon/by bus.’
b. Chelswu-ka ku chayk cecwuto-eyse sa-ass-ta.
-Nom the book at buy-past-M
‘Chelswu bought the book in Cecwu Island.’

Hong (1994) claims that the Caseless NPs in i) are Topics, therefore the data in i) are
not counterexamples to his generalization. However, it is difficult to accept his claim
because the Topic generally occupies the sentence initial position in Korean. On the other
hand, Hong’s claim induces a contradiction even if we admit it. Hong (1994) motivates
noun incorporation with the assumption that the Caseless N is an affix. If the Caseless
NPs are Topics in i), their head nouns must be affixes because the Topic marker is in
the complementary distribution with Case markers in Korean. Thus, we should make an
undesirable assumption that the Case-dropped N is an affix but the Topic-marker dropped
N is a free form.
192 Yong-Ha Kim

category is). Assuming that DPs with overt Case are in some non-
complement positions and have their Case feature checked off at overt
syntax while DPs with covert Case remain in the complement positions at
the overt syntax for the reason presented in 3.3. Then, we can preserve the
basic content of Hong’s Generalization.
To examine our interpretation of Hong’s Generalization, let’s consider the
contrast between verbs yeyppu- and philyoha-. Philyoha- in (15) might be
an unaccusative verb as Ahn (1996) claims, and nominative Case drop is
possible due to its seeming status as a (quasi-)complement. It is also likely
that yeyppu- the predicate of the embedded clause in (17b) is an unaccusative
verb. As an unaccusative verb, it should have a complement with the
nominative Case, and Hong’s Generalization predicts it is possible to drop the
nominative Case from its argument. Although yeyppu- and nwukwu in (17b) is
embedded in an ECM construction, their relationship is not changed and Exceptional
Case Marking is not obligatory in Korean as shown by the following sentences.

(18) a. Chelswu-ka Swunhi-ka yeyppu-ta-ko sayngkakha-n-ta.

b. Chelswu-ka Swunhi-lul yeyppu-ta-ko sayngkakha-n-ta.

Therefore, it’s not warranted to claim that the impossibility of Case drop in
(17b) is due to Exceptional Case marking. In fact, the Case drop from the
subject of yeyppu- is impossible even in a simple sentence. This
characteristic of yeyppu- contrasts to the property of philyoha-

*?
(19) a. Nwukwu yeyppu-ni?
b. Mwue philyoha-ni?

The contrast between the sentences in (19) strongly suggests that nwukwu in
 Overt Case and Covert Case in Korean 193

(19a) is not in the complement position. If so, yeyppu- is not unaccusative

though it superficially has only one argument. Then, what makes the two
sentences have different status with respect to the possibility of Case drop?
I think that the difference lies in the semantics of the two predicates in
(19).
Yeyppu- refers to a more or less permanent state, and might be an
individual-level predicate. Philyoha- refers to a temporary state, and might
be a stage-level predicate. That is to say, the difference between these two
verbs comes from Kratzer’s dichotomy of predicates. I believe that Diesing’s
(1992a) mapping principle can explain the contrast between the two
sentences in (19). According to Diesing (1992a), stage- and individual-level
predicates have different structures as follows.

(20) stage-level predicate

IP

Spec I
NP
I VP

Spec V
NP

(21) individual-level predicate

IP

Spec I
NPi
I VP

Spec V
PROi
194 Yong-Ha Kim

As represented in (20), Diesing claims that the subject of a stage-level

predicate is canonical in that it is base-generated in [SPEC, VP] and
subsequently moves to [SPEC, IP]. In contrast, the subject of an
individual-level predicate is base-generated in [SPEC, IP] and [SPEC, VP] is
occupied by a null pronominal element PRO which is controlled by the
subject. Diesing’s claim is that the structure of (20) allows the subject to
be interpreted as specific or generic in the IP-subject position, and
existential in the VP-subject position, whereas (21) has only one subject
position available, i.e. the IP-subject position, hence the indefinite subject
can be interpreted only as specific or generic. This is summarized in the
mapping principle below.

(22) Mapping Principle16

Material from VP is mapped into the nuclear scope.
Material from IP is mapped into a restrictive clause.

Diesing (1992a) further claims that in languages like English that requires
all subjects to appear in [SPEC, IP], the distinction between internal and
external subjects cannot be made at the overt syntax, whereas languages like
German allow VP-subject as well as IP-subject at the overt syntax, so the
stage/individual contrast can be expressed at the overt syntax. As for the
VP-subject at the overt syntax in German, she presents many pieces of
evidence, one of which is the so-called was-für split. In German, was für
functions as a specifier of an NP and can be split up by fronting was. But

16. The terms nuclear scope and restrictive clause are originated from Heim (1982). In
Heim’s (1982) theory, indefinites are treated as variables, and there are operators such as
abstract existential , every, some, not, and so on. and these operators unseletively bind
indefinites to result in a tripartite logical representation consisting of an operator, a
restrictive clause, and the nuclear scope. For the discussion and explanation of these
terms, See Heim 1982 and Diesing 1992a.
 Overt Case and Covert Case in Korean 195

the was-für split is applicable only when the NP of which was für is the
specifier is within VP.

*
(23) a. Was sind für Leguane intelligent?
what are for iguanas intelligent
‘What kind of iguanas are intelligent?’
b. Was sind für Leguane verfügbar?
available
‘What kind of iguanas are availble?’

(24) a. *Was sind für Abgottschlangen taub?

boa constrictors deaf
‘What kind of boa constrictors are deaf?’
b. Was sind für Abgottschlangen sichtbar?
visible
‘What kind of boa constrictors are visible?’

The predicates in the (b) sentences above are stage-level and the ones in
the (a) sentences are individual-level. Clearly, this contrast is due to the
fact that German stage-level predicates allow the subject to remain in VP at
the overt syntax whereas individual-predicates do not.
Korean shows a similar phenomenon to the German was-für split. Korean
has many numeral classifiers (NC) which consist of an adnominal numeral
and a classifier noun. They usually follow an NP and make an indefinite
expression. Furthermore, they can be overtly Case-marked like ordinary NPs.
I will call the NP + NC sequence as NCP and assume without argument
that in an NCP, the NC is the head and the NP it follows is its
complement, hence the entire expression made by the combination of an NC
with an NP is a kind of NP headed by the NC.17 As (25) shows, there are
196 Yong-Ha Kim

four different ways for arranging NCPs in Korean.

(25) a. Chelswu-ka haksayng han myeng-ul kaluchi-n-ta.

Chelswu-Nom student one Cl-Acc teach-Pres-M
‘Chelswu teaches a student.’
b. Chelswu-ka haksayng han myeng kaluchi-n-ta.
c. Chelswu-ka haksayng-ul han myeng-ul kaluchi-n-ta.
d. Chelswu-ka haksayng-ul han myeng kaluchi-n-ta.

In (25a) and (25b), the NP and NC normally constitute an NCP except

that the accusative Case marker is dropped in (25b). In (25c) and (25d),
they split up and make Double Object constructions, and in (25d) the NC
han myeng appears with the covert Case. If this kind of indefinite
expression appears as the subject of the stage- or individual-level predicate,
the contrast is strikingly parallel to the was-für split.

(26) a. Haksayng han myeng-i yeyppu-ta.

b. *Haksayng han myeng yeyppu-ta.
c. Haksayng-i han myeng-i yeyppu-ta.
d. *Haksayng-i han myeng yeyppu-ta.

(27) a. Haksayng han myeng philyoha-ta.

b. Haksayng han myeng-i philyoha-ta.
c. Haksayng-i han myeng-i philyoha-ta.
d. Haksayng-i han myeng philyoha-ta.

17. This assumption may seem awkward because the so-called NC is generally treated
as a quantifier which must be bound by an NP. But, the binding effect can be captured
under our assumptions because the NCP contains the trace of the complement NP when
it moves to some higher position.
 Overt Case and Covert Case in Korean 197

Suppose that the verb yeyppu- in (26) is an individual-level predicate and

philyoha- in (27) is a stage-level predicate, we can easily explain the
contrast by arguing that as in German, subjects can appear in either [SPEC,
IP] or some VP-internal position at the overt syntax in Korean, and NCs
with the covert Case occupy some VP-internal position. Which position,
then, do the NCs with the covert Case occupy in VP exactly? According to
Diesing (1992a), it is the Spec of VP. This is because Diesing (1992a)
accepts the VP-internal subject hypothesis and tacitly assumes that the
subject of a predicate always occupies the Spec of VP whether it is a
one-place predicate or not. I think, however, that there isn’t any forceful
reason to posit the subject of a one-place predicate always should be in
[SPEC, VP] in Korean. Thus, I will revise the structure in (20) and (21) as
follows for Korean predicates like philyoha- and yeyppu-:

(28) stage-level predicate

IP

Spec I
NP
VP I

NP V

(29) individual-level predicate

IP

Spec I
NPi
VP I

PROi V
198 Yong-Ha Kim

With this structure in mind, let’s consider the sentences in (27) first. In
(27a), the NCP headed by the NC can remain in the lower NP position in
(28) because Korean allows subjects to appear within VP by Spell-Out. In
(27b), the NCP can move from the VP-internal position to the Spec of IP
because Korean also allows subjects to appear in the Spec of IP at the
overt syntax. In (27c), the entire NP moves as in (27b), and in addition,
the complement NP in the NCP is separated and moves to another Spec of
IP, hence a Double Subject construction. In (27d), the NC remains in situ,
but the complement is separate and moves to the Spec of IP. All the
relevant representations of the sentences in (27) are as follows:

(30) a. [IP [VP [NCP [NP haksayng] han myeng] philyoha]]

b. [IP [NCP [NP haksayng] han myeng]i-i [VP ti philyoha]]
c. [IP haksayngj-i [NCP tj han myeng]i-i [VP ti philyoha]]
d. [IP haksayngj-i [VP [NCP tj han myeng] philyoha]]

In contrast, the ungrammatical sentences in (26) have a Case-dropped NC

which indicates that it must be in the lower NP position in VP that PRO
has already occupied. But, in the grammatical sentences in (26), the subjects
are in the Spec of IP that is the canonical position for the subject of
individual-level predicates.
Now, returning to the examples in (19) repeated here.

*?
(19) a. Nwukwu yeyppu-ni?
b. Mwue philyoha-ni?

If our analysis above is tenable, Hong’s generalization seems to not only

rule out (19a) but also rule in (19b). Yeyppu- in (19a) is an individual-level
 Overt Case and Covert Case in Korean 199

predicate and its subject is in the specifier position of IP, so that the Case
of its subject cannot be dropped, whereas philyoha- in (19b) is a stage-level
predicate and its subject can remain in the complement position of VP, so
the Case-drop is possible.
The discussion so far reveals that Hong’s generalization (with some
re-interpretation) successfully explains the aspects of Case drop in Korean.
However, we find the defect of Hong’s generalization if we apply it to the
subject of the transitive construction. Hong’s generalization predicts that the
subject of the transitive construction cannot be Case-dropped because it
cannot be in the complement position. But the following sentences show
this prediction is not fulfilled.

(31) a. Haksayng han myeng-i kkoch-ey mwul-ul cwu-n-ta.

student one Cl-Nom flower-Dat water-Acc give-Pres-M
‘A student waters the flowers.’
?
b. Haksayng han myeng kkoch-ey mwul-ul cwu-n-ta.
c. Haksayng-i han myeng-i kkoch-ey mwul-ul cwu-n-ta.
d. Haksayng-i han myeng kkoch-ey mwul-ul cwu-n-ta.

The NCP in (31) is the subject of a transitive verb and the sentences are
all grammatical.18 If we accept Hong’s Generalization, we should incorrectly

18. Though it’s not fully ungrammatical, (31b) is somewhat awkward in comparison
with the other sentences in (31). I don’t have a definite explanation for this. Notice that
the subject NCP with the covert Case is in the scope of the negation, whereas the NCP
subject with overt Case isn’t, as shown below.

i) Haksayng han myeng kkoch-ey mwul-ul cwu-ci an-h-nun-ta.

-Comp Neg-do-Pres-M
‘No student water the flowers.’
ii) Haksayng han myeng-i kkoch-ey mwul-ul cwu-ci an-h-nun-ta
200 Yong-Ha Kim

rule out (31b) and (31d). So, it seems that we must go the other way
around to account for the grammaticality of the examples in (31).
In fact, Hong (1994) presents some exceptions to his generalization. The
sentence (32) is one of them.

(32) Chelswu cengmal Yenghi-lul salangha-n-ta.

really -Acc love-Pres-M
‘Chelswu really loves Yenghi.’

Chelswu in (32) is the subject of the transitive verb salangha-, but the
nominative Case drop is nonetheless possible. Hong’s Generalization fails to
rule in the acceptable sentence (32). Hong’s (1994) solution to this apparent
problem is to treat the subject in (32) as the Topic of the sentence. It is
the same as Saito’s (1983) solution to the sentence (13) based on Kuno’s
(1973) claim that a wh-word cannot be the Topic. This claim is supported
by the sentence (17a) repeated below.

(17) a. *Nwukwu chayk-ul sa-ass-ni?

‘There is a student who does not water the flowers.’

The scope phenomena can be explained if we assume that the negation is above vP in
the sentence. On the other hand, sentence i) is much better than (31b), probably because
it contains the negation. If we change the mood of sentence (31b), its acceptability is
also much improved. Compare sentence iii) with (31b).

iii) Haksayng han myeng kkoch-ey mwul-ul cwu-lkes kath-ta.

Comp seem
‘It seems that a student will water the flowers.’

In any event, I don’t offer a satisfactory explanation for what is the factor of the
acceptability of the indefinite subject with covert Case.
 Overt Case and Covert Case in Korean 201

(17a) is ungrammatical because the impossible Case-drop from the subject

occurs. In order for nwukwu in (17a) to survive, it should be the Topic.
But this option does not work in this case because nwukwu is a wh-word.
So far, I have consciously evaded the discussion of (17a). Now, it’s time
to touch this example with respect to the discussion of the example in (32).
My position is that the subject in (32) is not the Topic and (17a) is bad
for an unknown reason. Thus, my arguments diverge from Hong’s
Generalization for the possibility of nominative Case drop.19
It seems generally accepted in general linguistics that the Topic is old or
given information (cf. Palmer 1981: 158-161). This primary property of the
Topic is confirmed by the following example.

(33) Speaker A: Nwu-ka nay chayk-ul ilk-ess-ni?20

Who-Nom my book-Acc read-Past-Q?
‘Who read my book?’
*
Speaker B: a. Chelswu-nun akka kuke-l ilk-ess-e-yo.
-Top ago it-Acc read-Past-Dec-Hon
b. Chelswu-ka akka kuke-l ilk-ess-eyo.
‘Chelswu read it a short time ago.’

19. A reviewer points out that (11a) is still worse than the sentence that contains the
overtly Case marked subject.

i) Enni ku chayk-ul ilk-nun-ta (=11a)

ii) Enni-ka ku chayk(-ul) ilk-nun-ta (=10b or 11b)

As mentioned in footnote 1, his judgement is quite different from the informants I

have consulted. In addition, i) should be grammatical if we agree with Hong (1994) on
the grammaticality of (32). Thus, the question seems to be how we can prove that the
Caseless subject of the transitive sentence is the subject not the Topic.
20. Nwukwu reduces to nwu when it is assigned the nominative Case.
202 Yong-Ha Kim

In (33), speaker A asks for new information about a person who read his
book. Speaker B’s answer (a) is unacceptable because the relevant
information is marked with the Topic marker -nun, whereas answer (b) is
acceptable because an NP with the nominative Case marker can serve as
new information. According to Hong’s claim, we may expect that the
sentence is unacceptable if the new information in Speaker B’s answer, that
is, the subject, appears with the covert Case. However, (34) shows that this
is not the case.

(34) Speaker A: Nwu-ka nay chayk-ul ilk-ess-ni?

Speaker B: Chelswu akka kuke-l ilk-ess-eyo.21

If Speaker B’s answer in (34) is acceptable, Chelswu is not the Topic. And
if it is not the Topic, it is a plain argument of the predicate ilk; that is,
the subject with the covert Case.
Let’s return to (31), which Hong’s Generalization fails to explain.
Suppose that the subject in sentence (31b) is the Topic, then we may
expect that the sentence is acceptable when we recover the Topic marker.
But this expectation fails again.

*?
(35) Haksayng han myeng-un kkoch-ey mwul-ul cwu-n-ta.

One might argue that (35) is acceptable and grammatical. Actually, (35) is
acceptable under the contrastive reading. However, for the sentence to be
acceptable, the indefiniteness of the subject must disappear. In other words,

21. The two reviewers mention that speaker B’s answer in (34) seems to be
inappropriate. As stated in footnote 18, the acceptability of the subject with covert Case
can be affected by the modality of the sentence containing it. Thus, speaker B’s answer
in (34) is appropriate for an evasive response.
 Overt Case and Covert Case in Korean 203

the entity the subject denotes must be definite. This claim is supported by
(36).

(36) *Haksayng-un han myeng kkoch-ey mwul-ul cwu-n-ta.

Thus, it is evident that the Topic and the subject with the covert Case have
different status, so that it is not reasonable to treat the subject with the
covert Case as the Topic.
There remains a problem concerning (17a) which our analysis might not
be able to handle. Consider the following data.

(37) a. Nwu-ka ku chayk-ul sa-ass-ni?

Who-Nom the book-Acc buy-Past-Q
‘Who bought the book?’
*
b. Nwukwu-nun ku chayk-ul sa-ass-ni?
Who-Top
*
c. Nwukwu ku chayk-ul sa-ass-ni?

The parallelism between nwukwu-nun and nwukwu in (37) strongly suggests

that the contrast in (37) be explained by the assumption that the subject
with the covert Case is the Topic. Though I do not yet have a clear-cut
explanation for the reason why the subject in (37c) cannot be a wh-word,
some comments are in order.
Sentence (37a) has another reading in which nwu-ka is not a wh-word
but an indefinite pronoun. Surprisingly, (37c) is grammatical if nwukwu is
interpreted as an indefinite pronoun, but nwukwu-nun in (37b) cannot be
used as an indefinite pronoun. The only way we save (37b) is to interpret
nwukwu-nun as a contrastive wh-word (cf. Han 1987: ch. 5)
204 Yong-Ha Kim

(38) a. Nwu-ka ku chayk-ul sa-ass-ni?

‘Did anybody buy the book?’
b. Nwukwu ku chayk-ul sa-ass-ni?
*
c. Nwukwu-nun ku chayk-ul sa-ass-ni?22

(39) Speaker A: Chelswu-nun acik ku chayk-ul an sa-ass-e.

yet not
‘Chelswu did not buy the book yet.’
Speaker B: a. Kulemyen, nwukwu-nun ku chayk-ul sa-ass-ni?
if so
‘Then, Who bought the book.’
*
b. Kulemyen, nwukwu ku chayk-ul sa-ass-ni?

What the data above suggest is clear; at least, nwukwu in (38b) and
Speaker B’s answer in (39) is not the Topic. Then, we can raise a question
immediately: Why do the contrasts in (37) on one hand and those in (38)
on the other ever exist? Though I don’t have a satisfactory answer to this
question as mentioned earlier, a descriptive analysis can be provided for the
contrasts. Based on the discussion so far, we may set up three positions for
subjects: a higher position for the Topic, the Spec of IP for the overtly
Case-marked subject, and a lower position for the subject with the covert
Case. Of these three positions, only the Spec of IP position allows
wh-words. Following Han (1987), let’s tentatively assume the position of the
Topic is the Spec of CP.23 What is the lower position for the subject with
the covert Case? Two candidates can be given. One is the Spec of vP and

22. As Han (1987) notes, (38c) can be grammatical if it has sarcastic implication. But
(38b) hardly have such implication and nwukwu-nun in this reading is not a Topic.
23. Objects can be the Topics, of course. I will ignore the Topicalization of objects
because the Topic is inserted into the current discussion only for the expository purpose.
 Overt Case and Covert Case in Korean 205

the other is the complement position of VP. Of the two positions, the Spec
of vP is more plausible because the complement position is generally for
objects.24 Let’s call the subject in the Spec of IP as IP-subject, and the
subject in the Spec of vP as vP-subject. The following diagram shows a
summary of the discussions in this section.25

24. Note, however, that the subject of an unaccusative verb with the covert Case is in
the complement position.
25. We are tempted from the discussion above to conclude hastily that wh-words
within vP or VP are illegal for some reason. However, we would immediately realize
that it is misled, facing data like (19b) repeated here.

i) Mwue philyoha-ni?

i) is an example to which Hong’s Generalization is applicable because mwue is the

complement of philyoha-, so it turns out that not all wh-words are illegal if they are
within vP or VP. Looking at more examples of this kind, we will learn that wh-words in
the complement position are legally interpreted although it appears with the covert Case.

ii) a. Chelswu-nun mwue sa-ass-ni?

buy-Past-Q
b. Ne nwukwu po-ass-ni?
you see-Past-Q

This is an unexpected subject-object asymmetry, and surely causes a problem in my

argument. I have no possible answer to this subject-object asymmetry at this stage, but
note that we have enough evidence to reject Case-dropped subject as Topic, so this
asymmetry should not be explained by the ban on wh-Topic.
206 Yong-Ha Kim

(40) Three positions for subjects

CP
wh-words impossible
specificity necessary Topic C'
Topic-marker drop impossible
IP C
wh-words possible
specificity necessary IP-subject I'
covert Case impossible
vP I
wh-words impossible
specificity impossible vP-subject v'
covert Case necessary
VP v

Complement V

3.2. Two positions for objects

From the discussions in the previous subsection, we can claim that the
overt Case indicates the subject is out of its thematic position, whereas the
covert Case indicates the subject is in its thematic position. Is it possible
for us to extend this claim to the objects? I believe so.
Enç (1991) reports that in Turkish, the specificity of indefinite
expressions is determined by the Case morphology: NPs with the overt Case
are necessarily specific whereas NPs with the covert Case are necessarily
non-specific.

(41) a. Ali bir piyano-yu kiralamak istiyor.

Ali one piano-Acc to-rent wants
‘Ali wants to rent a certain piano.’
b. Ali bir piyano kiralamak istiyor.
 Overt Case and Covert Case in Korean 207

‘Ali wants to rent a (non-specific) piano.’

Diesing (1992b) suggests a possible explanation for these data that the
specificity of Turkish object NPs is determined by the positions they
occupy. That is, an overtly Case-marked object moves from its base position
to a VP-external position and is mapped into a restrictive clause, hence the
specificity is obtained.
Mahajan (1990) also presents data in Hindi which show that overt object
agreement is related to the specificity.

(42) a. Raam-ne kitaab parhii.

Raam-ERG(M) book(F) read(PERF F SG)
‘Ram read the book.’
b. Raam-ne ek kitaab parhegaa.
Raam-ERG(M) a book(F) read(FUT M SG)
‘Ram will read a book.’

He argues, like Diesing (1992b), that only specific objects can move to the
Spec of AgrO as the object agreement morphology in (42a) shows. He,
therefore, suggests two ways of accusative Case-marking. Under his
suggestion, non-specific objects are assigned a Case under government by
verbs, but specific objects are assigned a Case from AgrO by SPEC-head
agreement.26
As their suggestions seem tenable, it is reasonable to apply it to Korean
data. Kim’s (1993) approach is exactly in accordance with their suggestions.

26. Mahajan (1990) suggests another way in which a Case is assigned under
government by AgrO. I will not take into account this way of Case marking. Readers
should refer Mahajan 1990 for this kind of Case assignment.
208 Yong-Ha Kim

Based on the discussions in Enç 1991 and Diesing 1992a, Kim (1993)
argues that Korean NPs show the specificity/non-specificity contrast with
respect to Case marking. See the following sentences.

(43) a. na-nun [sakwa sey kay-lul] mek-ess-ta.

I-Top apple 3 Cl-Acc eat-Past-M
‘I ate three apples.’
b. na-nun [sakwa sey kay] mek-ess-ta.

In (43a) the overtly Case-marked NCP sakwa sey kay-lul has a specific
reading, while the covertly Case-marked NCP sakwa sey kay in (43b)
doesn’t. Following Diesing’s (1992a) mapping principle that requires that
specific NPs be mapped into a restrictive clause (i.e. outside VP) and
non-specific NPs be mapped into the nuclear scope (i.e. inside VP), Kim
(1993) claims that the object NCP in (43a) moves to the Spec of AgrO at
the overt syntax, that in (43b) remain within VP, and they are assigned
overt accusative and abstract (covert) accusative respectively.27
An anonymous reviewer claims that it is possible to utter the sentence
(43a) even in case where sakwa sey kay-lul is not specific. I fully agree
with him on the judgement. The following examples confirm it.

(44) Speaker A: Ne daietu sicakha-ess-ni?

you diet begin-PAST-Q?
‘Have you begun to diet yourself?’
Speaker B: a. ung, kulayse halwu-e sakwa sey kay-lul mek-e.
yes so a day-in apple three Cl-Acc eat-Dec
‘Yes, I have eaten three apples a day, then.’

27. In Kim 1993, is introduced instead of AgrO as in Johnson 1991.

 Overt Case and Covert Case in Korean 209

*
b. ung, kulayse sakwa sey kay-nun halwu-e mak-e.
-Top

Sakwa sey kay-lul is not specific in Speaker B’s answer (a) because there
is no established set of apples. Speaker B’s answer (b) in which sakwa sey
kay-nun appears as the Topic is very deviant. Thus, it turns out, contra
Kim’s (1993) claim, that NCPs with the overt Case can appear with
non-specific reading.28
However, Kim’s observation is significant since the specificity/non-
specificity contrast between NCPs with and without the overt Case
nonetheless exists. Consider the examples below.

(45) a. Nay pang-ey ai-tul-i tule-o-ass-ta.

my room-Dat child-Pl-Nom. enter-come-Past-M
‘Children entered my room.’

28. The other reviewer says that it undermines the basis of the discrimination of the
overt Case and the covert Case if NCPs with the overt Case can be non-specific like
NCPs with the covert Case. But, it seems evident that NCPs with the covert Case cannot
be interpreted as specific as the discussion about (45)-(46) shows. In addition, if the
object of speaker B’s answer (a) in (44) has covert Case, it contrasts with the original
one.

i) ung, kulayse halwu-e sakwa sey kay-lul mek-e. (=speaker B’s answer (a))
ii) ung, kulayse halwu-e sakwa sey kay mek-e.

I think the contrast lies in the possibility of focus reading. The object with overt Case
in i) can receive focus reading, while the object in ii) cannot. It is well-established in
Korean linguistics that the accusative Case marker, as well as the nominative Case
marker, serves as a focus marker (see Y-H Kim 1988, G-H Kim 1988, and Schütze
1996). I will not tackle this issue because it will takes us too far away from our main
discussion.
210 Yong-Ha Kim

b. Na-nun sonye twu myeng-ul a-n-ta.

I-Top girl two Cl-Acc know-Pres-M
‘I know two girls.’
c. Na-nun sonye twu myeng a-n-ta

(46) a. Odam-a birkaç çocuk girdi.

my-room-Dat several child entered
‘Several children entered my room.’
b. Iki kiz-i kanïyordum.
two girl-Acc I-knew
‘I knew two girls.’
c. Iki kiz kanïyordum.

(46) is a hypothesized discourse in Turkish quoted from Enç 1991.

According to him, (46b) and (46c) are different in meaning. (46b) is about
two girls who are included in the set of children established by the
utterance of (46a). (46c) is about two girls who are excluded from the
original set of children. In other words, Iki kiz-i in (46b) is specific
whereas Iki kiz in (46c) is non-specific. (45) is the Korean counterpart to
(46), in which sonye twu myeng-ul in (45b) and sonye twu myeng in (45c)
can be equally interpreted as non-specific. However, for the ‘two girls’ to
be interpreted as specific in the discourse, (45b) must be selected. We can
conclude from these examples that indefinite NPs must be overtly
Case-marked in Korean in order to be interpreted as specific. If we try to
capture this difference syntactically, the most plausible way is to accept the
proposal in Mahajan 1990 and Diesing 1992b and slightly modify it to the
effect that overtly Case marked objects are in a VP-external position, so
mapped into a restrictive clause.29
 Overt Case and Covert Case in Korean 211

Now, we should determine the VP-external position for overtly Case

marked objects. Since the Case-checking replaces the Case-marking or
Case-assignment in the minimalist program, we may select the Spec of vP
as the VP-external position. It appears, however, that we are in a
predicament because the subject with the covert Case can also occupy the
Spec of vP. Thus, we are in a situation to say that elements in the same
configuration can be differently interpreted with respect to specificity. To
escape from this predicament, let’s see the insight hidden in our discussions:
overtly Case marked DPs are all in checking positions, whereas DPs with
the covert Case are all in thematic positions. Assuming that checking
position and thematic position replace the VP and IP in Diesing’s (1992a)
mapping principle respectively, we can modify it as follows.

(47) Mapping Principl e30

29. A reviewer points out that it is dubious that the specificity effect that overt and
covert Case show can be used as evidence for the movement of the object. Unlike
Turkish, Korean allows a definite expression or a proper name to appear with the covert
Case (the Turkish example from Dede 1986)

i) Na-nun onul Doli po-ass-ta. (Korean)

I-Top today see-Past-M
‘I saw Doli today.’
ii) *Bugün Aytül gördüm. (Turkish)
today see-Past-1SG
‘Today, I saw Aytül.’

Of course, proper names and definite expressions are inherently specific. Turkish never
allows proper names and definite expressions to appear with the covert Case, so the
specificity effect of Case markers is clear. I do not have a plausible explanation for this
difference between the two languages. However, the contrast in (45) seems significant, so
that we can say that Korean also exhibits the specificity effect of Case markers.
30. If we assume Agr-based Case theory as in Watanabe 1993 and the clausal
architecture in Koizumi 1995, we can easily apply Tsai’s (1994) extended mapping
212 Yong-Ha Kim

Material from a thematic position is mapped into the nuclear scope.

Material from a checking position is mapped into a restrictive
clause.

Though (47) may have a potential problem, let’s be satisfied with it

tentatively. (48) is an extended summary of our discussions so far.31

(48) Positions for DPs

CP
wh-words impossible
specificity necessary Topic C
Topic-marker drop impossible
IP C
wh-words possible
specificity possible IP-subject I
covert Case impossible
vP I
wh-words impossible
specificity impossible vP-subject v
covert Case necessary

wh-words possible vP-object v

specificity possible
covert Case impossible VP v

wh-words possible Complement V

specificity impossible
covert Case necessary

hypothesis to our analysis. Tsai (1994) assumes a cyclic algorithm for the mapping
principle.
31. A reviewer raises the question whether there is further evidence of the two
positions for the object. I have no further evidence for it at present, but the discussion in
Hong 1994 can be used as indirect evidence at least in the case of the object if we
agree on the re-interpretation of Hong’s Generalization by the checking theory.
 Overt Case and Covert Case in Korean 213

3.3. Determination of the strong feature

Hitherto, I have left the categorial status of Case markers undetermined.

In this subsection, I will suggest that the data in question can be well
explained by assuming that the Case markers are overt Ds in Korean. Since
Ahn 1988, Case markers in Korean have been often treated as belonging to
the category D (cf. Hong 1994, Yang 1996). Let’s take a look at the
arguments for the analysis of Case as D.
The first argument is that we can conjecture that the counterpart to D in
head-initial languages like English will be the rightmost element of the
nominal construction in a typical head-final language like Korean. The
Korean Case marker is the most plausible element because it is the
rightmost element of the nominal construction.
Second, though they are affixes, Korean Case markers make syntactic
constituents with the phrase preceding them. This means that they are not
morphological affixes but a syntactic unit attached to phrases.
Third, Korean Case markers have to do with the specificity effects as
discussed earlier in much detail. The specificity is a typical property of the
category D, thus it is highly plausible to analyze Korean Case markers as
Ds.
Following this line of reasoning in Ahn 1988, let’s assume that Case
markers belong to the category of D. And further assume that they have
Case features in addition to the categorial feature [D]. Then, nominative
Case marker -i/ka is a D with the nominative Case feature, and -ul/lul is a
D with the accusative Case feature.32 We could extend this assumption to

32. One can claim that it is undesirable to call -i/ka and ul/lul as Case markers, if
their features are divided into the categorial feature [D] and the Case features. I think
this claim is right, though I will continue to use the term Case marker for the sake of
convenience.
214 Yong-Ha Kim

the covert Case to the effect that there are phonetically null Ds having
Case features in Korean. We will see that this assumption reveals important
consequences.
Consider the formal features operative at the computation in (6), repeated
here.

(6) a. categorial features

b. ø-features
c. Case feature
d. strong F, where F is categorial.

During the selection of lexical items from the Lexicon to the numeration,
the features in (6) are also introduced with the lexical items. Note that the
strong feature belongs only to the categorial features. The features in (6a-c),
thus, are the necessary features of lexical items, but the strong feature is
associated only with attractors parametrically. Suppose this is the case, then
how can we determine whether an attractor has the strong feature or not?
Chomsky (1995: 294f) suggests the following economy principle.

(49) enters the numeration only if it has an effect on output.

The output is the PF representation or the LF representation. (49) means

that if a lexical item or its relevant features are introduced into the
numeration, they must have an effect at PF or LF. This actually amounts to
saying that a certain element can be inserted after Spell-Out if it doesn’t
have any semantic content or any features interpreted at LF, and that an
element can also be inserted into LF if it has only features interpreted at
LF but not any phonetic features. The strong feature is a feature which
 Overt Case and Covert Case in Korean 215

must be checked off as soon as it is introduced into a derivation, otherwise

the derivation crashes. Because of this property, the strong feature is
operative only at the overt syntax; that is, it forces a derivation to violate
Procrastinate. If the strong feature enters the numeration, it has an effect
only on PF.
To see how (50) can explain syntactic phenomena, let’s consider sentence
(50) below (Chomsky 1995: 292ff).

(50) Who knows John?

Sentences like (50) cause a problem in English, which commonly is

assumed to have the interrogative C with the strong wh-feature. If the
English interrogative C has the strong wh-feature as assumed, (50) should
be the following form.

*
(51) Who does know John?

As the asterisk indicates, English speakers never take (51) to be

grammatical. Chomsky (1986) claims that learners of English establish a
grammar without the movement of subject wh-phrases if they have accessed
only sentences like (50). This claim is equal to the economy principle (49),
in that even if English has a parameter of the interrogative C with the
strong wh-feature, it never enters the numeration in the case of the
wh-subject construction because there is no effect of the strong feature at
PF.
If the strong feature causes the overt movement, how much material can
move at the overt syntax? Suppose that I in (52) has the strong D-feature,
then it will attract Subj to check it off at the overt syntax.
216 Yong-Ha Kim

(52) [ I [ [Subj John’s book] v ....

As mentioned in section 2, every movement is essentially a feature movement

under the minimalist assumption. It is desirable for the formal feature [D] in
Subj to move alone. However, the scattered features are not subject to PF-rules
according to Chomsky (1995), so all the other features including phonetic features,
Case features, and so on should carry along with [D]. In this case, the
minimal material to be moved is the head of Subj, that is, ’s. This movement
of ’s alone will be also barred because there must be a filter like (53) at PF.

(53) Affixes cannot be stranded.

Therefore, the host John should go along with ’s. But, John’s is not a
syntactic object as shown in (54), and hence not subject to movement.

(54) DP

John D

’s book

It seems that the whole DP should move for the PF-convergence, so that
the overt movement demands the whole category to move. Chomsky (1995)
posits the following economy condition for this reason.

(55) F carries along just enough material for convergence.

With the discussions in mind, let’s turn to Korean in which we concern

ourselves. We have seen that the IP-subject and vP-object necessarily require
 Overt Case and Covert Case in Korean 217

the overt Case while the vP-subject and VP-object necessarily require the
covert Case. Thus, it is likely that the overt Case is checked at the overt
syntax and the covert Case is checked at the covert syntax.33 In fact, it is
the claim I make in this paper. This means that I and v have the strong
D-feature in Korean only when the overt Case markers appear. Some
natural questions, however, arise immediately; Why do I and v have the
strong feature only if the subject and object have the overt Case markers?
Why are I’s and v’s strong features reflected on DPs?
Suppose that Korean I has the strong D-feature and there is a subject
with the overt nominative Case marker.34 It must attract the categorial
feature [D] in the subject. Under our assumption, it belongs to the
nominative Case marker -i/ka. It is, then, desirable to raise only -i/ka for
the checking of the strong D-feature of I. However, the Korean Case
markers are affixes, so that the filter in (53) and the economy condition
(55) demands that the whole DP move. This is shown in (56) below.

(56) a. *-kai Chelswu-ti o-ass-ta.

Nom Chelswu come-Past-M
‘Chelswu came.’
b. [Chelswu-ka]i ti o-ass-ta.

33. A reviewer makes an objection against the distinction between overt Case and
covert Case. His point seems to be that the distinction is not justifiable if both the
expressions with overt Case and covert Case have the same Case feature to be checked at
the overt syntax or LF. However, I do not think of overt and covert Case as distinct
categories of Case. Rather, I think there is a null D in Korean, which is an allomorph of
the overt D and has the same Case feature as the overt D. Thus overt Case and covert
Case are cover-terms expressing the existence and non-existence of Case markers. As the
reviewer points out, overt Case just means the Case checked at the overt component and
covert Case the Case checked at the covert component.
34. For convenience’s sake, I only take I(nfl) into account for the moment. But, the
discussion will be easily extended to v.
218 Yong-Ha Kim

What happens in a sentence containing the strong I and a subject with

the covert nominative Case marker? Is it necessary for the whole subject
DP to move? It is natural to expect that the covert nominative Case marker
is not subject to the filter in (53) due to its lack of phonetic features, so
that the movement of the whole DP is not tolerated by the economy
principle (55). Thus, we may conclude that the covert Case marker raises
alone. However, there is no output effect if the covert Case marker raises
alone:

(57) a. -Øi Chelswu-ti o-ass-ta.35

Chelswu come-Past-M
b. [Chelswu-Ø]i ti o-ass-ta.

Recall that the strong feature’s entrance into the numeration is blocked
by the economy principle (49); it results in a vacuous movement of the
covert Case marker as shown above, hence no PF effect. Thus we can
reasonably conclude that when the subject DP (and the object DP) has the
covert Case marker as its head, I cannot have the strong D-feature even if
its basic parameter is strong, like the English C in a sentence containing
wh-subject.36

35. Ø stands for the covert Case marker.

36. A reviewer asks how we can block the overt movement of DPs with covert Case
when I’s or v’s D-feature is strong, and also asks how to make DPs with the overt Case
move when I’s or v’s D-feature is not strong. I think of I’s and v’s D-feature as
basically strong in Korean. But recall that the economy principle (49) demands that the
strong feature have some output effects. Even if the phonologically null Case marker is
attracted by some functional head and moves to the checking domain, its movement has
no PF effects. Therefore, I’s D-feature, as well as v’s, cannot be strong by (49) when
DPs with the covert Case appear, and, by the same token, DPs with the overt Case are
always attracted by I or v and move to the Spec of I or v because the basic feature
strength cannot be canceled without some appropriate reason.
 Overt Case and Covert Case in Korean 219

One might argue that (56b) can be derived with the subject remaining in
situ because the subject can maintain its sentence-initial position. I think
this possibility is natural under our assumption, but data concerning
scrambling in Korean suggest that this possibility cannot be implemented.
Let’s consider the examples in (58).

*
(58) a. Cakii emeni-ka Chelswu-luli salangha-n-ta.
self mother-Nom -Acc love-Pres-M
‘Hisi mother loves Chelswui.’
b. Chelswu-luli cakii emeni-ka salangha-n-ta.

Although there have been many disputes on the nature of scrambling, the
contrast between the examples in (58) strongly suggests that scrambling has
the properties of A-movement.37 If we accept the simplified clause-structure
in Chomsky 1995, the most plausible position for the scrambled object
Chelswu-lul is the Spec of IP because it is the only possible A-position
higher than its thematic position.38 Suppose that the subject raises first in
(58b). The formal feature [D] and the Case feature can enter into a
checking relation with I. Chomsky (1995) proposes two effects of checking;
deletion and erasure. Deletion means being invisible at LF but accessible to
the computation, and erasure means being inaccessible to any operation.
Suppose further that I’s strong D-feature can escape erasure (i.e. it is just
deleted), so that it can attract the object. The formal feature [D] of the

37. It is generally accepted in work in generative grammar that scrambling has dual
character of A- and A’-movement (see Mahajan 1990, Webelhuth 1995, Saito 1989, Lee
1994, Park 1994 among others). I will restrict the discussion to A-scrambling.
38. In fact, there is another possibility in (58): the subject is in the inner Spec of v,
and the object in the outer Spec of v. This possibility, however, is blocked independently.
See the following subsection.
220 Yong-Ha Kim

object is interpretable, and is accessible to the operations at the computation

whether or not it once checked v’s D-feature as we have discussed in
section 2. Thus, it can check off the strong D-feature of I. The derivation
of (58b) is legitimate. Note that the escape from erasure is not helpful in
the former case because I’s strong D-feature is not visible at LF even
though it is just deleted. It seems, then, that the strong feature of I should
be posited for independent reason.39
Another question can be raised about our assumptions: How can we
determine which Case the subject and object have when it appears with the
covert Case? Consider the following examples:

(59) a. Swunhi Chelswu cohaha-n-ta.

like-Pres-M
not ‘Chelswu likes Swunhi.’
but ‘Swunhi likes Chelswu.’
b. Chelswu Swunhi cohaha-n-ta.
not ‘Swunhi likes Chelswu.’
but ‘Chelswu likes Swunhi.’

As the translations indicate, (59b) cannot treated as a scrambled version of

(59a), and vice versa. (59a) and (59b) are correctly interpreted by most
Korean speakers, so they are equally grammatical with different meanings.
If we cannot determine the Case features of the arguments, it is possible

39. In the first draft of this paper, I postulated feature strength for attractees, so that
overt movement can take place only when both the attractor and the attractee have strong
features. The two anonymous reviewers criticize this postulation simultaneously because it
is not compatible with Chomsky’s (1995) theoretical framework. The above discussion can
be a possible solution to this problem. See Lee 1996 for the consequences of the
postulation of attractees’ feature strength concerning verb movement in English.
 Overt Case and Covert Case in Korean 221

that the subjects and the objects in (59) have the same Case features, and
hence cannot derive the two sentences because we cannot determine which
sentence blocks the other sentence: Economy consideration is possible only
among convergent derivations in a reference set with the same numeration.
But if we can determine the Case features of the arguments in (59a-b), we
can derive both sentences because they are derived form different
numerations.
We can conclude from the discussions in this section that the overt Case
marker indicates the strong D-feature of I and v, but the covert Case
marker indicates the lack of the strong D-feature in I and v.

3.4. Analysis of Data

Now, let’s consider how our assumptions work with the main data in
(10)-(12) repeated here.

(10) a. Enni ku chayk ilk-nun-ta.

b. Enni-ka ku chayk-ul ilk-nun-ta.

(11) a. Enni ku chayk-ul ilk-nun-ta.

b. Enni-ka ku chayk ilk-nun-ta.

*
(12) a. Ku chayk enni ilk-nun-ta.
b. Ku chayk-ul enni-ka ilk-nun-ta.
c. *Ku chayk-ul enni ilk-nun-ta.
*
d. Ku chayk Enni-ka ilk-nun-ta.

Let’s start with the derivation of (10b). At some point in the derivation, we
222 Yong-Ha Kim

might merge the light verb v with VP in a transitive sentence.40

(60) vP

VP v

Obj V

With v having the strong D-feature signified by the overt Case, we can
take one of the two possible next steps. In one of the two alternatives,
Subj enni-ka is merged with the vP, and then, v attracts Obj ku chayk-ul to
its Spec to form a new vP, and Obj enters into a checking relation with v.
The result is that Subj is in the inner Spec of v and Obj in the outer Spec
of v, as in (61).41

(61) vP

SPEC2 v
Obj
SPEC1 v
Subj
VP v

tObj V

In the other alternative, v attracts Obj in the first place. Subj, in turn, is

40) Following Chomsky (1995), I will use Subj for subject and Obj for object
respectively.
41. I will ignore the issue of the timing of the verb movement in Korean. It is likely
that my framework equally explains the data whether or not the verb movement takes
place overtly.
 Overt Case and Covert Case in Korean 223

merged with the new vP. The result is the reverse of (61): the subject is in
the outer Spec of v and the object in the inner Spec of v, as in (62).

(62) vP

SPEC2 v
Subj
SPEC1 v
Obj
VP v

tObj V

Chomsky (1995) prefers (61) to (62) without clear reasons, but we will, or
must, select (62) rather than (61) for a few reasons. One of them42 is that,
as Chomsky (1995) notes, we can eliminate the notion of equidistance (63),
to simplify the notion closer than as in (65), and reject (64); in (63)-(64),
c-commands and in (65), is not in the minimal domain of the
target K.

(63) and are equidistant from if and are in the same

minimal domain.

(64) is closer to the target K than unless is in the same

minimal domain as .

(65) is closer to the target K than if c-commands .

42. The other reasons are empirical in nature. We can explain the word order
variation in (11)-(12) only when we choose (63) over (62). See below.
224 Yong-Ha Kim

If we assume the simpler notion of closer than in (65) and select the
structure (62) over (61), Subj always c-commands Obj, and then, only Subj
can be attracted by I. Hence, (62) is simpler and more plausible than (61).
On the other hand, Dr. Dong-Whee Yang (personal communication) points
out that (61) is more economical than (62) because Subj is introduced by
Merge, whereas Obj is affected by Move. Although it is assumed in
Chomsky 1995 that Merge is more economical than Move, Ura (1996)
argues that there is a reason for the selection of (62) over (61). If v has
the strong D-feature as we have assumed in Korean, it must disappear
through checking before further operations extend the structure. If Subj is
merged into v’s Spec earlier than Obj, the derivation crashes because the
merged Subj cannot enter into any checking relation at that position;43 v’s
strong D-feature does not disappear. If Obj moves first and checks off the
strong feature of v, then the derivation can survive and continue further. It
is an apparent violation of the economy condition, but the only way to save
the derivation is to move Obj before Subj is introduced.44 This is what the
Last Resort Condition requires (See Ura 1996 for a detailed discussion).
Furthermore, (61) is completely incompatible with the sentences (11a) and
(12c) repeated here again.

(11) a. Enni ku chayk-ul ilk-nun-ta.

(12) c. *Ku chayk-ul enni ilk-nun-ta.

43. Notice that the position of Subj is in a thematic position and thematic positions
cannot be checking positions. If thematic positions can be checking positions it will allow
illegal movement to thematic positions. See Chomsky (1994).
44. Recently, Collins (1997) has claimed that Merge and Move are equally costly, so
that (61) and (62) should not be discriminated by the economy considerations.
 Overt Case and Covert Case in Korean 225

If DPs with the covert Case remain in their Ɵ-positions as discussed

above and we choose (61) rather than (62), (11a) should be ungrammatical
and (12c) should be grammatical. However, the actual judgement is quite
the reverse. To cover the sentences (11a) and (12c), we must select (62)
over (61). This point is clarified below.
In (62), the formal feature [D] in Obj checks off v’s strong D-feature.45
Then, Subj moves to the Spec of I containing the strong D-feature, and
simultaneously checks off I’s strong D-feature and its Case feature in (66).

(66) IP

Spec I
Subj
vP I

SPEC2 v
tSubj
SPEC1 v
Obj
VP v

tObj V

Let’s consider the derivation of (10a) which is a bit more complicated

than the derivation of (10b). Subj and Obj remain in situ, i.e. Subj in the
Spec of v and Obj in the complement position of V. At the overt syntax,

45. The checking of the accusative Case is between V and Obj. If V-movement
occurs at the overt syntax in Korean, the checking of the accusative Case also occurs
there with strong v, otherwise it occurs at the covert syntax. As mentioned earlier, this
does not seem to affect our discussions, so I will leave it open. It should be noted,
however, that the nominative Case checking is between I and Subj, thus the checking of
nominative Case as well as EPP feature, occurs at the overt syntax if I is strong.
226 Yong-Ha Kim

there are no operations because (10a) doesn’t have any element with the
strong D-feature as the DPs with the covert Case indicate. At the covert
component, I should attract FF(Subj) first and then v should attract FF(Obj)
after it moves to I.46 The reason why the movement of FF(Subj) and
FF(Obj) should be ordered is that there is a condition which requires that
intermediate traces not prevent the attraction of an element that it
c-commands (Chomsky 1995: 301-304):

(67) Only the head of a chain enters into the operation Attract/Move.

If FF(Obj) is attracted by v and moves prior to the movement of FF(Subj),

it goes across Subj violating the MLC (Minimal Link Condition). To the
contrary, if FF(Subj), attracted by I, moves earlier than FF(Obj), it leaves
its trace in the inner Spec of v. FF(Obj) is, in turn, attracted by v and
moves across the trace of Subj without violating the MLC because
condition (67) says that the trace of Subj is not accounted to be a barrier
to the movement of FF(Obj). The structure of (10a) at LF is (68) for the
relevant parts.47

46. If v moves to I at the overt syntax with its D-feature unchecked, v-to-I movement
should take place earlier than the movement of FF(Obj). Again, this does not affect our
analyses. See Chomsky (1995).
47. Chomsky (1995) wants to eliminate the structure like (68). The reason is that he
thinks it is a natural requirement with respect to output condition that FF(Subj) must
0max
c-command FF(Obj) in I if both are present. Hence, he makes the conclusion that at
least one of them must raise overtly. However, I do not agree with Chomsky (1995)
because the requirement which he supposes to be the case is not logically derived. When
Obj A-scrambles across Subj containing some anaphor in languages like Korean and
Japanese, it can bind the anaphor in Subj as discussed in 3.3.

i) *Caki-uyi emeni-ka Chelswu-luli salangha-n-ta.

self-Gen mother-Nom -Acc love-Pres-M
 Overt Case and Covert Case in Korean 227

(68) I0max

FF(Obj) I

FF(Subj) I

v I

V v

The derivation of (11a) and (11b) repeated here is as simple as (10b).

(11) a. Enni ku chayk-ul ilk-nun-ta.

b. Enni-ka ku chayk ilk-nun-ta.

In (11a), v has the strong D-feature, and hence attracts Obj. The formal
feature [D] in Obj checks off the strong D-feature. As the covert Case of
Subj suggests, I appears without the strong D-feature, so that covertly,
FF(Subj) is attracted by I and enters into a checking relation with I. (69a-b)
are the overt and covert representations of (11a) respectively.

ii) Chelswu-luli caki-uyi emeni-ka salangha-n-ta.

If the binding conditions apply at LF as Chomsky (1992) assumes, the object Chelswu-lul
must c-command caki-uy. Thus, I just ignore the requirement that FF(Subj) must
0max
c-command FF(Obj) in I .
228 Yong-Ha Kim

(69) a. IP b. IP

vP I vP I

SPEC2 v tFF(Subj) v' FF(Subj) I

Subj
SPEC1 v Obj v
Obj
VP v VP v

tObj V tObj V

In (11b), Obj is frozen in its Ɵ-position at the overt syntax. Subj is

overtly attracted for the feature-checking by I containing the strong
D-feature. At the covert component, FF(Obj) is attracted by the sublabel v
under I, and then, enters into a checking relation with v. The relevant
structures are (70a-b).

(70) a. IP b. IP

Spec I Subj I

vP I vP I

Tsubj v tSubj v FF(Obj) I

VP v VP tv v I

Obj V tFF(obj)

Now, let’s figure out the derivation of the sentences in (12) repeated here
again, which generally are pretty complex.
 Overt Case and Covert Case in Korean 229

*
(12) a. Ku chayk enni ilk-nun-ta.
b. Ku chayk-ul enni-ka ilk-nun-ta.
c. *Ku chayk-ul enni ilk-nun-ta.
*
d. Ku chayk Enni-ka ilk-nun-ta.48

In (12a), both Subj and Obj enter the numeration with the covert Case,
so that it is expected that both of them are in their Ɵ-positions. However,
(12a) shows the reverse order of (10a) in which Subj and Obj are also
expected to be in their Ɵ-position.49 Because lowering is prohibited

48. The two reviewers point out that (12d) is a well-formed sentence (cf. Yi 1988). In
fact, (12d) seems to be grammatical at first glance. I believe, however, that (12d) is an
instance of the left dislocation if it is accepted as grammatical. As the examples below
show, the object in (12d) licences a resumptive pronoun, while it cannot if it is
topicalized or scrambled.

i) a. Ku chayk, Enni-ka Ø ilk-nun-ta.

b. Ku chayk, Enni-ka kuke-l ilk-nun-ta.
it-Acc
ii) a. Ku chayk-un Enni-ka ilk-nun-ta.
-Top
b. *Ku chayk-un Enni-ka kuke-l ilk-nun-ta.
iii) a. Ku chayk-ul Enni-ka ilk-nun-ta.
-Acc
b. *Ku chayk-ul Enni-ka ku ke-l ilk-nun-ta.

The problem with these data seems to be whether i a) can be admitted as a left
dislocation construction, which does not have the overt resumptive pronoun. Since Ross
1967, the left dislocation has been treated as different from the topicalization (see Lasnik
& Saito 1992) for it has the resumptive pronoun. But, Ø in i a) can be regarded as a
null resumptive pronoun since Korean allows null object (cf. Kim 1995), and hence i a)
is an instance of the left dislocation.
49. If ku chayk is Subj and enni is Obj in (12a), the sentence might be grammatical
with the meaning ‘the book reads my sister.’ But the book (chayk) does not have the
ability to read (ilk-) something, the sister (enni) is not an entity which can be read, so
230 Yong-Ha Kim

conventionally, the only way to derive (12a) is to move Obj across Subj,
and the only possible landing site for Obj in (12a) is the outer Spec of v
because the movement of Obj to the Spec of I violates MLC in virtue of
intervening Subj which is closer to the target than Obj. However, Obj in
(12a) cannot move there because we have selected (62) rather than (61) to
simplify the notion closer than.
We may produce (12b) through scrambling of Obj over Subj from (10b).
Suppose that Korean may have a parametric option that the strong D-feature
is allowed to escape erasure though being checked, as we have assumed in
the discussion of (58). Scrambling (at least A-scrambling) takes place when
I exercises this option. In (12b), Obj is attracted by v, and checks off the
strong D-feature of v. I subsequently attracts Subj, and checks off the
strong D-feature and the nominative Case against it, but the strong
D-feature is only deleted this time. Hence, I still has an accessible (i.e.
active) strong feature, so that it can attract Obj and check it off against
Obj. The result is a multiple Spec construction.

(71) [IP Obj [I' Subj [I' [vP tSubj [v' tObj [v' [VP tObj V] v]]] I]]]

In (12c), the movement of Obj ku chayk-ul is wrong. If (12c) is derived

by scrambling, the derivation is as follows. Obj firstly moves to the inner
Spec of v, Subj is merged into the outer Spec of v, and then I is merged
with vP containing two Specs. Finally, Obj moves to the Spec of I. Here,
we meet a problem. The overt movement of Obj to the Spec of I means
that I has the strong D-feature. Does the overt Case of Obj signify I’s
strong D-feature? In other words, can I attract the formal feature [D] in
Obj despite the covert Case of Subj? If the answers to these questions are

that even in this case, the sentence is gibberish.

 Overt Case and Covert Case in Korean 231

negative, the overt movement of Obj to the Spec of I in (12c) has no

motivation, thus is an illegal movement. If the answers to these questions
are affirmative, the overt movement of Obj to the Spec of I in (12c)
violates MLC because of the intervening Subj that is closer to I than Obj
in the simplified sense of closer to; hence (12c) crashes.

(72) [IP Obj [I' [vP Subj [v' tObj [v' [VP tObj V] v]]] I]]
MLC

Whichever answer we may choose, the derivation of (12c) is illegal.50

Finally, the ungrammaticality of (12d) results from the illegal movement
of Obj which has the covert Case and hence should not be affected by the
overt operations. Even if the strong D-feature in I can escape erasure after
Subj checks it off in (12d), it can only attract the formal feature [D] in the
covert Case marker vacuously at the overt syntax, the pied-ping of the
whole Obj DP is a violation of the economy principle (55). The movement
of Obj in (12d), hence, has no motivation.

(73) [IP Ø [I' Subj [I' [vP tSubj [v' [v' [VP Obj-tØ V] v]]] I]]
no PF effect

50. I will not consider any further which answer is the right one because there seem
to be many complicated matters involved. I leave it for future research.
232 Yong-Ha Kim

4. Conclusion

In this paper, I have proven that the contrast between overt and covert
Case in Korean can be well explained under the assumptions (i) that Case
markers belong to the category of D and hence have the formal categorial
feature [D] as well as the Case features, and (ii) that there are phonetically
null Case markers in Korean, which also have the formal feature [D] and
the Case features. Based on these assumptions and the discussions in the
above sections, we can answer the questions given in section 1 of this
paper which are repeated here.

(3) a. Why is word order rigid in Korean when every argument NP has
no Case marker?
b. Why is word order free in Korean when every argument NP has
a Case marker?
c. Why cannot free word order appear even in case that only one
of the argument NPs has no Case marker?

The answer to (3a) is that when an NP appears with the overt Case
marker, it is not influenced by any grammatical operation at the overt
syntax because its null phonetic content causes vacuous movement and
invalidates the effect of the strong feature at PF. Thus, the word order is
rigid in Korean when every argument NP has no Case marker.
The answer to (3b) is that when an NP appears with the overt Case
marker, it signifies the strong D-feature of the relevant functional element,
and hence is ready to be influenced by any grammatical operation at the
overt syntax for convergence. Thus, the word order is free in Korean when
every argument NP has a Case marker.
 Overt Case and Covert Case in Korean 233

The answer to (3c) is that when one of the argument NPs has the covert
Case marker, it is not ready to be influenced by any grammatical operation
and does create a kind of minimality barrier to any A-movement which
crosses over it. Hence, the word order is rigid.
Though it seems that there are many issues unanswered under our
proposals, I hope that future research can extricate them.

References

Ahn, Hee-Don. 1988. Preliminary remarks on Korean NP. In Papers from

the Sixth International Conference on Korean Linguistics, ed.
Eung-Jin Baek, 1-15, Seoul: Hanshin Publishing Co.
Ahn, Hee-Don. 1996. A study of syntactic word-formation. In Morphosyntax
in Generative Grammar, ed. Hee-Don Ahn, Myung-Yoon Kang,
Yong-Suck Kim & Sookhee Lee, 1-62, Seoul: Hankuk Publishing Co.
Ahn, Byung-Hee. 1966. For the establishment of Casus Indefinitus. Tong-A
Mwunhwa 6. [Republished in Modern Korean Grammar, ed. Ki-Shim
Nam, Young-Keun Koh & Iksop Lee, 99-101, Taegu: Keimyung
University Press (1975).]
Baker, Mark C. 1988. Incorporation. Chicago: The University of Chicago
Press.
Chomsky, Noam. 1957. Syntactic Structure. Hague: Mouton.
Chomsky, Noam. 1986. Barriers. Cambridge, Mass.: MIT Press.
Chomsky, Noam. 1991. Some notes on economy of derivation and representation.
In Principles and Parameters in Comparative Grammar, ed. Robert
Freidin, 417-454, Cambridge, Mass.: MIT Press.
Chomsky, Noam. 1992. A minimalist program for linguistic theory. MIT
234 Yong-Ha Kim

Occasional Papers in Linguistics 1. MITWPL, Department of

Linguistics and Philosophy, MIT, Cambridge, Mass. [Published in The
View from Building 20: Essays in Linguistics in Honor of Syllvain
Bromberger, ed. Kenneth Hale & Jay Keyser, 1-52, Cambridge,
Mass.: MIT Press (1993).]
Chomsky, Noam. 1994. Bare phrase structure. MIT Occasional Papers in
Linguistics 5. MITWPL, Department of Linguistics and Philosophy,
MIT, Cambridge, Mass. [Published in Government and Binding
Theory and the Minimalist Program, ed. Gert Webelhuth, 383-439.
Oxford: Blackwell (1995).]
Chomsky, Noam. 1995. The Minimalist Program. Cambridge, Mass.: MIT
Press.
Collins, Chris. 1997. Local Economy. Cambridge, Mass.: MIT Press.
Dede, Müserref. 1986. Definiteness and referentiality in Turkish verbal
sentences. In Studies in Turkish Linguistics, ed. Dan Isaac Slobin & Karl
Zimmer, 147-163, Amsterdam: John Benjamins Publishing Company.
Diesing, Molly. 1992a. Bare plural subjects and derivation of logical
representation. Linguistic Inquiry 23:353-380.
Diesing, Molly. 1992b. Indefinites. Cambridge, Mass.: MIT Press.
Enç, Mürvet. 1991. The semantics of specificity. Linguistic Inquiry 23:1-25.
Fukui, Naoki & Margaret Speas. 1986. Specifiers and projection. In MIT
Working Papers in Linguistics 8, 128-172, MITWPL, Department of
Linguistics and Philosophy, MIT, Cambridge, Mass.
Han, Hak-Sung. 1987. The configurational structure of the Korean language.
Doctoral Dissertation, The University of Texas at Austin.
Heim, Irene. 1982. The semantics of definite and indefinite noun phrases.
Doctoral Dissertation, University of Massachusetts Amherst.
Hong, Yong-Tcheol. 1994. Incorporation theory and the distribution of Case
 Overt Case and Covert Case in Korean 235

morphemes. Studies in Generative Grammar 4:1-43.

Johnson, Kyle. 1991. Object positions. Natural Language and Linguistic
Theory 9:577-636.
Jonas, Dianne & Jonathan D. Bobaljik. 1993. Specs for subjects: The role
of TP in Icelandic. In MIT Working Papers in Linguistics 18: Papers
on Case and Agreement, 59-98, MITWPL, Department of Linguistics
and Philosophy, MIT, Cambridge, Mass.
Kim, Dae-Bin. 1993. The Specificity/Non-Specificity Distinction and Scrambling
Theory. Seoul: Thaehaksa.
Kim, Gwi-Hwa. 1988. A study on Case of the Korean Language. Master’s
Thesis, Sogang University.
Kim, Yong-Suck. 1995. On null objects in Korean. Korean Journal of
Linguistics 20:1-37
Kim, Young-Hee. 1988. Exploration in Korean Syntax. Seoul: Top Press.
Kim, Young-Hee. 1991. Conditions on unmarked Case. In Journal of Language
Studies 9, 5-32, Language Institute, Keimyung University, Taegu.
Kim, Young-Joo. 1990. The syntax and semantics of Korean Case; The
interaction between lexical and syntactic levels of representation.
Doctoral Dissertation, Harvard University.
Kim, Young-Joo. 1991. Nominative Case marking in Korean. Studies in
Generative Grammar 2:113-148.
Koizumi, Masatoshi. 1995. Phrase structure in minimalist syntax. Doctoral
Dissertation, MIT, Cambridge, Mass.
Kuno, Susumu. 1973. The Structure of Japanese language. Cambridge,
Mass.: MIT Press.
Lee, Jeong-Shik. 1996. Verbal inflection: A lexicalist-minimalist approach.
Korean Journal of Linguistics 21:483-498.
Lee, Rangheyoon K. 1994. Economy of representation. Doctoral Dissertation,
236 Yong-Ha Kim

University of Connecticut.
Mahajan, Anoop Kumar. 1990. The A/A-bar distinction and movement
theory. Doctoral Dissertation, MIT.
Palmer, F. R. 1981. Semantics. Cambridge: Cambridge University Press.
Park, Soon-Hyuck. 1994. A- and A-bar dependencies and minimalism in
movement theories: Scrambling and binding. Doctoral Dissertation,
University of Wisconsin, Madison.
Pollock, Jean Y. 1989. Verb movement, Universal Grammar and the
structure of IP. Linguistic Inquiry 20:365-424.
Ross, John R. 1967. Contraints on variables in Syntax. Doctoral
Dissertation, MIT, Cambridge, Mass.
Saito, Mamoru. 1983. Case and government in Japanese. In Proceedings of
the West Coast Conference on Formal Linguistics 2, 247-259. Stanford
Linguistics Association, Department of Linguistics, Stanford University,
Stanford, CA.
Saito, Mamoru. 1985. Some asymmetries in Japanese and their theoretical
implication. Doctoral Dissertation, MIT, Cambridge, Mass.
Saito, Mamoru. 1989. Scrambling as semantically vacuous A'-movement. In
Alternative Conceptions of Phrase Structure ed. Mark R. Baltin &
Antony S. Kroch, 182-200. Chicago: Chicago University Press.
Saito, Mamoru & Howard Lasnik. 1992. Move : Conditions on Its
Application and Output. Cambridge, Mass.: MIT Press.
Shütze, Carson T. 1996. Korean “Case stacking” isn’t: unifying noncase
uses of Case particles. In Proceedings of the North East Linguistic
Society 26, 351-365. Reproduced and distributed by GLSA, Department
of Linguistics, University of Massachusetts, Amherst, Mass.
Tsai, Wei-tien Dylan. 1994. On economizing the theory of A-bar dependencies.
Doctoral Dissertation, MIT, Cambridge, Mass.
 Overt Case and Covert Case in Korean 237

Ura, Hiroyuki. 1996. Multiple feature-checking: A theory of grammatical

function splitting. Doctoral Dissertation, MIT, Cambridge, Mass.
Watanabe, Akira. 1993. AGR-based Case theory and its interaction with the
A-bar system. Doctoral Dissertation, MIT.
Webelhuth, Gert. 1995. X-bar theory and Case theory. In Government and
Binding Theory and the Minimalist Program, ed. Gert Webelhuth,
17-95, Cambridge, Mass.: Basil Blackwell.
Yang, Dong-Whee. 1994. The Theory of Grammar. Seoul: Hankook
Publishing Co.
Yang, Dong-Whee. 1995. Lectures on the minimalist program. Ms., The
Linguistic Society of Korea, Seoul.
Yang, Dong-Whee. 1996. The Korean Case structure in the minimalist
program. Korean Journal of Linguistics 21:139-189.
Yi, Nam-Soon. 1988. A Study on the indefinite Case and Case-Marker
Ellipsis in Modern Korean. Seoul: Top Press.
Yim, Young-Jae. 1983. Case assignment and insertion of delimiters in
Korean. In Language and Linguistics 10, 83-96, Language Institute,
Hankuk University of Foreign Studies, Seoul.
Yim, Young-Jae. 1984. Case tropism. Doctoral Dissertation, University of
Washington.