427 Nominal Functional Categories in Korean: A Comparative Study with Languages with DP* 28 Mi-Jeung Jo (Pukyong University) The goal of this paper is to determine major functional categories concerning NPs in Korean and to characterize them in comparison with determiners in articled languages. D in languages with DP crucially has syntactic and semantic roles for NPs in such a language; it functions as the subordinator which turns them into an argument, and it has the ability to pick out a single instance of whatever described by an NP. In addition, D is the locus of a lexically-governed semantic nominalizing operator which shifts a property realized in an adjective into a generic and kind reference. It will be argued that the phrasal Case Marker adjacent to the phrase-final head is the most feasible subordinator of an NP in Korean. The head in complement-head construction in Korean is the locus of nominalizing operator; a grammatical noun kes can productively nominalize a clause in all possible positions of an NP. An articless language like Korean appears to have a syntactic nominalizing operator, which is due to the two parametric factors, the head-final parameter and phrasal morphological Case. 0. Introduction The parallelism between the internal structures of a noun phrase and a * This paper is written under the financial support of Pukyong University, made in the program year of 2000. 428 Mi-Jeung Jo sentencce extends to functional categories related to these constituents and their syntactic roles. A number of independent proposals have suggested that noun phrases containing one or more functional head are parallel to COM, INFL and AGR in the sentential system (cf. Abney (1987), Reuland (1983), Ritter (1991), Szabolcsi (1987, 1994)). The purpose of this paper is to characterize two nominal functional categories in Korean. The first topic concerns the question of determining the functional superstructure of NPs which subordinates them as arguments. The second topic is about a syntactic realization of a nominalization operator in Korean which is distinguished from that in languages with articles. In Section 1.1, we will introduce a theory on functions of determiners and parametric variations concerning them. Section 1.2 shows that an articless language such as Chinese can have an actual and an abstract subordinator for a definite and an indefinite NP, despite the fact that a bare N(P) is freely allowed to occur. In Section 1.3, it will be argued that the parallelism between functions of complementizers and determiners must be expanded to their positions in determining the subordinator of NPs in Korean; we propose that the phrasal Case-markers in Korean appearing in NP-final position is the most feasible candidate. Section 2 compares two different loci of a nominalization operator in an articled language such as Greek and a non-articled language such as Korean. 1.0. Subordinator of an NP in Korean Postulation of DP headed by functional category Determiner over a noun phrase has been generally accepted as necessary in languages with articles. It is, however, less clear whether a functional category corresponding to it exists in languages without articles. The aim of this section is to find a nominal functional category in Korean which has a function similar to determiners via two steps. I will first examine the functions of determiners in detail, and then Nominal Functional Categories in Korean: ... 429 discuss the functional superstructures of NPs in Chinese, which does not have overt determiners like Korean. 1.1. Two Functions of Determiners: Subordinator and Demonstrative Determiners are assumed to have two separate functions regarding an NP. Szabolcsi (1987, 1994:181) and Stowell (1989) claim that only DPs can function as arguments; NPs are predicates or “propositions,” and D turns them into arguments. Szabolcsi considers that in this respect typical Ds like articles are similar to complementizers which subordinate a clause. In addition, D has an individualizing or singularizing function (Longobardi 1994:634); it has the ability to pick out a single instance of whatever is described by an NP. This is the function of a quantifier/demonstrative, labeled as “the deitic function of D” by Cheng and Sybesma (1999). The two distinct functions of determiners is originated from Bhatt and Yoon (1992)'s observations on two distinct functions of complementizers; they argue that items broadly classified cross-linguistically as complementizers both serve as subordinator and indicate clause type. In languages like English these two functions are lexicalized in a single morpheme. In many languages with rich agglutinative morphology, however, these are carried by two separate morphemes. Korean, Japanese and Hungarian are classified as such languages. Szabolcsi (1994:218) cites such examples from Korean as in (1): (1) a. Bill-i [John-i wa-ss ta]-ko sayngkakha-n-ta Nom Nom come-past-dec-Comp think-pres-dec ‘Bill thinks that John came.’ b. Bill-i [John-i wa-ss nunya]-ko mul-ess-ta 430 Mi-Jeung Jo Nom Nom come-past-Q-Comp ask-past-dec ‘Bill asked whether John came.’ Szabolcsi further claims that the two functions of determiners can be either conflated or lexicalized separately, analogous to the typology of complementizers. English is a language which typically conflates these two functions as in the case of complementizers, whereas Hungarian systematically lexicalizes them as separate morphemes; both an article and other determiners may cooccur, as in (2a), and a complementizer and a Wh-word indicating interrogative sentence can appear together, as in (2b): (2) a. az é n minden allitas-om the I(Nom) every claim-poss.1sg b. Nem tudom, hogy hol not ‘my every claim’ van Já nos know-I SUB where is John(-Nom) ‘I do not know where Janos is.’ Under her analysis, Hungarian represents a typical language which has a pure subordinator both at the clausal and the noun phrase level. If Szabolcsi is correct on her insight on the parallelism between the functions of determiners and complementizers and the parameters set on these functions, it seems plausible to pose a question concerning an articless language such as Korean; which element undertakes the function of the subordinator for an NP argument? Szabolcsi (1994:213) is inclined to think that Korean has a construction similar to the Hungarian example in (2a), citing the data in (3): Nominal Functional Categories in Korean: ... (3) i/ku motun salam this/that all 431 ‘all the(se) people’ people Nevertheless, the construction in (3) is misinterpreted, even under her generalization on the two separate functions of determiners; i /ku ‘this/that’ are not determiners equivalent to the article in Hungarian az in (2a); they are demonstratives appearing within the structure of an NP. The structural identity of the demonstratives in Korean will become clear in a construction in which it interacts with a possessive pronoun or NP. Both a determiner and a possessive can cooccur in Hungarian (cf. (2a) and Italian, as in (4). In Italian, the first element appearing on the left of the nominal head is always the article, and it also precedes the possessive pronoun (Giorgi and Longobardi 1991:203): (4) a. il mio libre (the my book) b. *mio il libre (my the book) However the relative order between a demonstrative and a possessive pronoun in Korean is in reverse to that of the Italian and Hungarian counterparts (cf. (2)): (5) a. *ku na-ui the I-Gen chaek (lit. the my book) book b. na-ui ku chaek (lit. the my book) I-Gen the book Therefore, we assume that the structure of an NP in Korean concerning a 432 Mi-Jeung Jo demonstrative is similar to (6a), rather than (6b)1 : (6) a. [N max [DP D ] [NMAX-1 . . . N . . . ]] b. [DP D [NMAX . . . N . . .]] Since a demonstrative in Korean ku is an optional element appearing within an NP it is inappropriate to postulate DP as the superstructure of NPs functioning as their subordinator.2 Fukui (1986:Ch. 4) has a similar observation on Japanese demonstratives such as ko-no ‘this’ and a-no ‘that’; he observes that they behave like English prenominal modifiers and that they do not have a property of closing off the category projection, as in (7) (ibid:202): (7) a. ko-no hon b. John-no ko-no hon ‘this book’ lit. ‘John's this book’ c. akai John's ko-no hon lit. ‘red John's this book’ On the bases of this, he concludes that Japanese lacks the functional category DET and hence, that noun phrases in Japanese are a projection of N, N' which are always “open”. Although NPs in Korean, like the Japanese counterparts, are not “closed off” 1. The two different structures of an NP in (5) is labeled as ‘the NP parameter’ (5a) and ‘the DP parameter’, respectively (Jo M.-J. (1999)). 2. Since the demonstratives i/ku/ce can modify only nouns, they are posited as a specifier of an NP. In the head-last parameter, the domain of specifiers and that of complement are not strictly divided as in the head-first parameter. Therefore, movement of these determiners close to the head must be independently posited. Nevertheless, they cannot occupy in the head position of DP which is structurally superior and precedes a possessive NP. Nominal Functional Categories in Korean: ... 433 by the functional category DET, it does not necessarily mean that no functional category plays the role of the subordinator of an NP which can close off the projection of N. Therefore, I will pursue the question of which functional category is the subordinator of NPs in Korean in section 1.3. In the next section, the NP structures in the two Chinese languages, Mandarin and Cantonese, and their subordinators proposed by Cheng and Sybesma (1999) will be introduced in order to show that the NPs in these articless languages also can have an overt and an abstract superstructure for an indefinite and a definte NP. 1.2. Existence of an Abstract Subordinator for Bare Nouns The obligatoriness of DP for an NP argument forces one to postulate an empty D in languages with an article system. Longobardi (1994)'s proposal for the distribution of bare nouns in some Romance languages and Germanic languages can be summarized as follows. The core generalization is that bare nouns with an indefinite interpretation are restricted to lexically governed positions (essentially, object position). Longobardi claims that bare nouns are not really bare: they are embedded in a full-fledged DP structure, with an empty D head. The presence of an empty D head leads to an explanation for both the indefinite interpretation and the restricted distribution of bare nouns. However, there are other interpretations of bare nouns, involving generics and proper names, that are not restricted to lexically governed positions. Longobardi accounts for such bare nouns by arguing that the N-to-D movement has taken place, either in overt syntax or at LF. Thus, the empty D-node of DP appears to have a syntactic role of restricting occurrences and interpretations of bare indefinite nouns in these languages. 434 Mi-Jeung Jo Chinese is an articless language, which allows bare nouns to appear as arguments, and it has no distinction between mass/count nouns in terms of plural morphology. According to Chiechia(1998)'s Nominal Mapping Parameter, NPs in a language like Chinese are inherently argument, containing features [-predicate, +argument], and hence they need not be licensed by a functional category such as D. However, Cheng and Sybesma (1999), basically adopting Szabolcsi's theoretical claim, discuss the restricted distribution of bare nouns interpreted as indefinite and the issue of licensing them in Mandarin and Cantonese. They show that bare nouns in Chinese must involve more structure than just the bare N (or the bare NP). Bare nouns in both Mandarin and Cantonese can have indefinite and generic reading; only in the former language bare nouns have definite reading, whereas in the latter language, only [classifier+N] (i.e., [Cl + N])) can be interpreted as definite. Bare nouns with an indefinite reading are restricted to post-verbal position in both languages. Since noun phrases with overt numerals can only be interpreted as indefinite in the two languages, they further assume that indefinite bare noun and [Cl + N] phrases are in fact, NumeralPs with an empty Numeral head/ Cl head or only an empty Cl, analogous to the empty D in DP as proposed by Longobardi.3 Under this account, surface strings of the form [Cl + N] have two different structural representations; [Cl + N] with an indefinite reading is a NumeralP, as in (8a), and [Cl + N] with a definite reading is a ClP, as in (8b). 3. The interpretational possibilities of Bare nouns and [Cl + N] phrases in Cantonese(C) and Mandarin(M) are summarized in the following table (Cheng and Sybesma (1999:512): Indef Def Generic Bare nouns M/C M M/C [Cl + N] M/C C ____ Nominal Functional Categories in Korean: ... (8) a. NumeralP Numeral b. ClP Cl 435 ClP Cl NP NP N N As for bare nouns, the one with an indefinite reading has the structure of (8a), and the one with a definite reading has the structure of (8b) with an empty functional head(s). To summarize, bare nouns in the two Chinese languages are not really bare. Cheng and Sybesma conclude that the classifier in Chinese expresses both classification and number, at the same time performing the deictic/subordinative function; thus the two languages have the conflated form of the two functions of determiners as English, according to the typology that Szabolcsi proposed. The analysis of Chinese NP by Cheng and Sybesma shows that functional categories other than an article, such as numerals and classifiers, can function as an overt and an abstract subordinator for an NP argument in an articleless language. Turning to the question of NPs in Korean, it also has numerals accompanying with classifiers for both count and mass nouns(, similar to Mandarin).4 Nevertheless, it is unclear whether interpretation of an indefinite bare N(P) is structurally governed, and thus, there is no motivation for postulating an abstract subordinator for a bare N(P). 4. Cheng and Saysma claim that this phenomenon is related to the lack of plural/number morphology in Chinese. This claim seems to hold in Korean, a language without number morphology. 436 Mi-Jeung Jo 1.3. The Phrasal Case-Marker as the head of KP In this section, we will argue that position of a functional category must be taken into consideration in determining the subordinator of NP in Korean; unlike Chinese, Korean has phrasal morphological Case markers adjacent to the head, and they not only marks Case of an NP but also they subordinate it as an arguments. The location of the subordinators of an NP and a clause appear to be a significant factor for the typology of these functional categories; position of a determiner, or other elements equivalent to it, is parallel to that of a complementizer. They will be either phrase/clause-initial or phrase/clause-final. As for the Hungarian examples in (2), determiner az and complementizer hogy are both in the initial position of an NP and a clause, respectively. Similarly, the determiner and the complementizer in English also occupy the initial position.5 A complementizer in Korean is in clause-final and a morpheme representing a sentence type is adjacent to it, as shown in the sentences in (1). This fact suggests that the subordinator of an NP argument must be positioned in NP-final. Szabolcsi disregards the fact that there exists no parallelism between the positions of a complementizer and a demonstrative in Korean, as shown in the examples in (1) and (3) in which the former is in the clause-final and the latter, in phrase-initial. The most feasible candidate for the subordinator of an 5. Since the subordinator of an NP, overt or abstract one, is in the phrase-initial position in Chinese, as given in the structures in (6) we can assume that a complementizer must be in the clause-initial position. However, we are informed that Chinese has no overt complementizer for an embedded clause. The absence of an overt complemtizer does not counter the parallelism in the positions of the two subordinators. Nominal Functional Categories in Korean: ... 437 NP is the phrasal Case markers such as ka/i for the Nominatives and (l)ul for the Accusatives; they accompany all NP arguments regardless of whether they are definite and indefinite. In contrast, (n)un, which is known as the Topic marker in the initial position of a sentence can appear with both an NP and a postpositional phrase (PP). Therefore, we can see that the two Case markers function as the subordinators of an NP argument which turn it into an argument, assigning Case at the same time. We postulate K (Case) in KP as the superstructure of NP in Korean as in (9): (9) KP NP K N ka/i (Nominative) (l)ul (Accusative) The Nominative marker attached to a predicate NP (or a noun which is a part of a predicate) is the default Case marker for a nonargument, as the underlined elements in the sentences in (8)6 : (10) a. Suil-i Nom uisa-ka toe-ess-ta doctor-Part become-Past-Dec ‘Suil has become a doctor.’ b. Sunae-ka Nom maum-i coh-0-ta heart-Part good-Pres-Dec ‘Sunae is good-hearted.’ The nonarguments followed by the default Case marker in (10) behave 6. Stowell (1989) pointed out that the predicate phrase (of nominal Small Clause) are nonreferential cateogires and a bare noun is allowed in such a context, as in we elected John as president of the class. 438 Mi-Jeung Jo differently from an argument with the Nominative/Accusative Case marker; the former cannot be freely reordered, as in (11), while the latter can be freely reordered, as in (12). (11) a. *uisa-ka Suil-i toe-ess-ta doctor-Part Nom b. *maum-i Sunae-ka heart-Part become-Past-Dec coh-0-ta Nom good-Pres-Dec (12) a. Suil-i Sunae-eyke senmuwl-ul cu-ess-ta Nom to present-Acc give-Past-Dec ‘Suil gave a present to Sunae’ b. senmuwl-ul Sunae-eyke Suil-i cuw-ess-ta The sentences in (12) show that a Case marker attached to an argument allows free reorderings. However, such a property is absent in a default Case marker given to a nonargument. The existence of a grammatical element which functions as a subordinator of an NP as well as a Case-marker is not an idiosyncratic phenomenon. Among languages with DP, determiners in German carry out these two functions, in addition to quantifying an NP (i.e., definiteness). K in KP in Korean, on the other hand, is the subordinator of NPs, which is independent from the ‘deitic function’ of D in a language with articles. The two functions of K in Korean seem to be parallel to those of the complementizer -ko (cf. (1b); it is a subordinator of a clause, and it is also assumed to be the Case-assigner of a clausal argument (cf.Yim Y.-J. (1984)). Hungarian has a morphological Case as well as the article system. The Nominative Case is unmarked, and it is explicitly represented in a verb, and Nominal Functional Categories in Korean: ... 439 only the Accusative Case is morphologically marked, as in (13) (an example slightly modified from Szabolcsi (1994:222)): (13) (En) eltitkolI om a /ezen/ valamenyi talalkozas-t keep secret-Def.1SG the/this/each meeting- ACC ‘I keep[def] the/this/each meeting secret.’ Given that the two functions of D are realized in separate morphemes, the morphological Case in Hungarian (Zero or marked) represents a pure Casemarker To conclude Section.1, it appears that languages without an article system chooses a language-specific subordinator of NPs among various nominal functional categories. In Chinese, Numeral P and ClP are the functional categories which subordinate indefinite and definite NPs, respectively, and the heads of these categories are allowed to be empty in restricted syntactic contexts for specific readings. Although Korean shares significant common properties concerning NP with Chinese, such as the lack of articles and plural morphology (i.e. classifiers for both mass and count nouns) it has, unlike Chinese, a morphological Case system; phrasal Case-markers such as ka/i (Nominative) and (l)ul (Accusative) carry out the role of NP subordinator in Korean. They can be omitted only in colloquial Korean as long as a sentence maintains the basic word order (i.e. SOV order). 2. The Semantic Nominalization versus The Syntactic Nominalization Operator Having the different NP structures presented in (6), or having a different 440 Mi-Jeung Jo functional category as the NP sudordinator seems to have far-reaching consequences to other phenomena relating to the structure of an NP. One of them involves the locus of a nominalization operator. This section will discuss two different loci of a nominalization operator in a language with articles such as Greek and a language without them such as Korean. In the former the head of DP, D involves nominalization of adjectives interpreted as generic and kind references. In contrast, the head of NP, N in the latter functions as a clausal nominalizer. The former process is a lexically governed semantic nominalization, but the latter process involves a productive syntactic nominalization. It is well-known that the ‘expletive’ definite determiner appears in generic DPs in articled languages. For example, English has a curious form of adjective nominalization, which can be used for generic and kind reference as in (14): (14) a. The rich are greedy. b. The poor are common. The nominalized adjectives trigger plural agreement on the predicate and must be accompanied by the definite article. Chierchia (1998) assumes that there is a null nominal head with a plural feature in the syntactic structure of these constructions as in the follwoing: (15) DP D NP ADJ N [+pl] the rich △ Nominal Functional Categories in Korean: ... 441 However, the question is why the resulting structure in (15) cannot be shifted to denoting a kind without resorting to the definite article. Chierchia provides an unsatisfactory account; phonologically null items must be licensed, but English lacks null determiners; therefore, the presence of a null head with nominalized adjectives can be licensed in English only by an overt determiner. This cannot account for why the overt determiner can license a null head only for some adjectives (e.g. *the addicted). Giannakidou and Stavrou (1999) discuss the construction similar to those in (14) in Greek as an instance of substantivization as in the following sentence: (16) I plusii the rich.MASC.PL.NOM ksexnun apo pu ksekinisan forget.3PL from where started.3PL ‘The rich forget where they started from.’ They clearly distinguish this phenomenon from another DP-internal phenomenon in Greek, subdeletion, which involves ellipis; they treat a bare adjective in a construction like (16) as a kind-denoting nominalization and the definite determiner, as a type-shifting device. In other words, in their analysis, no empty head is involved in the Greek counterpart, and the definite determiner is the syntactic instantiation of the nominalization operator. Giannakidou and Stavrou admit that the Greek pattern of substantivization is not exactly identical to the relevant English (or Dutch) pattern. They argue that unlike the English counterpart, the nominalized adjective in a construction like (16) is a noun, not an adjective. Evidence for this are a fixed gender for the nominalized adjective (contra various gender agreement of regular adjectives with the head N) and exclusion of adverb modification as well as unavailability 442 Mi-Jeung Jo of comparatives and superlatives.7 The semantics of substantivization, as they point out, should account for the four characteristics of this construction:(1) the obligatory presence of the definite article, (2) its kind-denoting, generic semantics, (3) the resistance to modification for adjectives and (4) lexical constraints on the derivation of the construction. The use of the definite article in Greek has two basic domains, namely the referential use for definite description and the generic use for generic DPs. In the referential use, the definite determiner denotes the extensional iota-operator; in the generic use, it denotes the intensionalized version of iota-operator. Giannakidou and Stavrou (1999) treat generic readings as the product of nominalization, in the sense of Chierchia (1998), and consequently handle the definite article in this case as the syntactic realization of the nominalization operator. In their analysis, the definite article in generic DPs is not at all expletive(as proposed in Longobardi (1994)), but contentful: the definite determiner in substantivization construction, such as the sentence in (16), plays the role of a nominalization operator. Chierchia (1998)'s nominalization involves mapping between properties and 7. Sentence (14) in the text with a different gender will be ungrammatical as in (1a), and no degree adverb or comparative/superlative form is allowed in this construction, as shown in (1b) and (1c), respectively: (1) a. *I plusies ksexnun apo pu ksekinisan the rich.FEM.PL.NOM FORGET.3PL. from where started.3PL. b. *I poli anapiri griniazoun sinexos the very handicapped complain.3PL constantly ‘The very handicapped constantly complain.’ c. *I plusioteri ksexnun apo pu ksekinisan the rich.COMPR forget.3PL from where started.3PL ‘The richer forget where they started from.’ Nominal Functional Categories in Korean: ... 443 kinds; to any natural property, say DOG, there corresponds a kind, say dog-kind. And kinds are intensionalized individuals; dog-kind can be thought as the individual concept that comprises the totality of all dogs in all possible worlds. Generally, there are type-shifting functions that allow us to express the correspondence between kinds and properties and get from the one to the other. One such function is the nominalization function ∩. If DOG is the property of being a dog, then ∩DOG is the corresponding kind. Thus, an adjective representing a certain property can be directly intensionalized by the definite determiner, and the adjectives nominalized by it resist modification for adjectives. Giannakidou and Stavrou's analysis based on the concept of Chierchia's nominalization provides an adequate account for the three semantic properties of the substantivization, except for the last one. Nevertheless, not all adjectives in Greek can be nominalized by the definite determiner, as with the English counterparts. Except for natural kinds, not all properties correspond to kinds and what constitutes a kind is not settled in the grammar; Giannakidou and Stavrou regard that it depends on the shared knowledge of the speaker community. Due to the unsettled semantics of the nominalization process, substantivization in Greek is subject to the lexical constraints. Likewise, the variation across languages can be explained: i eksartimeni ‘the addicted’ is a successful nominalization in Greek but the addicted is ungrammatical in English. Hence, in principle, the inventory of substantivizations is not expected to be identical across languages. The final semantic property of the substantivization, the lexical constraint on the nominalized adjective is adequately accounted for. To summarize this section, the definite determiner in generic DPs is the locus of the semantic nominalization operator in a language in which D is the 444 Mi-Jeung Jo subordinator of NPs, such as Greek. This shows that the definite determiner appearing in D has an additional syntactic function other than the two functions we have discussed in Section 1.1. It is unlikely that the substantivization construction [definite determiner + adjective] is available in a language like Korean, which has a demonstrative semantically equivalent to the definite determiner, ku. In the following, however, we will show that a grammatical noun appearing in the head position of the noun complement structure functions as a syntactic nominalization operator. There is a large class of nouns in Korean which are termed as ‘dependent’ or ‘defective’ nouns, since they have to be obligatorily expanded as a noun phrase. Jo M.-J.(1999) regards this as a characteristic phenomenon in an articless language (or a language with the structure of NPs in (6a)); it further differentiates such nouns depending on whether they fulfill syntactic properties of NPs with respect to their internal structure and their distributions in a sentence. Among them, noun kes, represented as [N, THING], appears to be nothing defective syntactically as a noun; it can be modified by specifiers of N such as i/ku/ce/etten (this/the/that/which), and it can occur in all possible syntactic positions for an NP. Although its semantic content is minimal, as represented in a single feature, [N, THING] it is consistently maintained in all these positions. Noun kes, furthermore, does play several morphosyntactic roles from which ordinary nouns and other ‘dependent’ nouns are excluded; it can be a nominalizer of a clause in addition to other functional roles.8 9 The syntactic 8. Kes is also a part of inflectional elements such as the future tense complex -ul kes iand the conjecture modal complex ss-ul kes i-. 9. Therefore, it is treated as a ‘grammatical noun’ which is the nominal counterpart of a ‘grammatical verb’ such as ha ‘do’. Nominal Functional Categories in Korean: ... 445 contexts for these grammatical roles can be formalized as in (17))10 : (17) If kes in the head position of an NP has no argument status in its complement clause (and an S), then it will undertake a certain grammatical role. The minimal semantic content of kes is neutralized precisely in the contexts described in (17). A construction like (18a) meets the condition in (17), where noun kes bears no relation to an argument in the preceding clause; it functions as a clausal nominalizer, being totally devoid of meaning. Noun kes in a construction like (18b), however, is related to an argument(i.e., coindexed with an empaty NP) in a relative clause as are other relativized NPs11 , and thus, this construction is excluded from the context described in (17); it retains its semantic content, [THING]12 : (18) a. Sunae-nun [S Suil-i kot TM ttena-l] kes(*sasil)-ulo Nom soon leave-Adn(future) thing(fact)-as po-n-ta see-Pres-Dec ‘Sunae regards that Suil will leave soon.’ 10. It is the revised version of (14) in (Jo, M.-J.(1999):460). 11. This is not the restriction particularly limited to kes, but the general one to all relativized NP. Thus, a predicate NP, which is a non-arguement cannot be relativized: (1) * [S Sunae-ka 0i toe-n] uisai 12. Yang D.-W. (1972:133) and Fukui and Speas (1987:135) deal with a relative NP headed by a pronoun in Korean and Japanese, repectively. The latters presents the relative NP headed by a pro-form such as sore ‘it’ in order to show that such a pro-form is not ‘closed off’ as the English counterpart. However, the relative NP headed by -kes is different from that headed by a pro-form in Korean such as ku ‘he’, ku-kes (lit. the-thing) ‘it’; it can independently occur in an S, while -kes should be modified by a complement clause or other elements. 446 Mi-Jeung Jo b. [S Sunae-ka 0i coha-ha-nun] Nom kesi-i mues-i-0-nya? good-do-Adn(Pres) thing-Nom what-is-Pres-Q 'What is the thing(s) that Sunae likes?' Noun kes cannot be replaced by a noun which selects a clausal complement such as sasil ‘fact,’ as in (18a), and therefore, it is not a proform of a certain noun. It is a noun which behaves like a functional category even in the head position of NP. Given the explicitly pronounced context for its grammatical roles in (17), we can treat kes as a legitimate nominalizer for clauses (and VPs/APs) like the nominalizing affixes in Korean -um and -ki. However, Yoon, H.-S.(1989: 213-5) excludes the noun-complement structure like that in (18a) from nominalization in Korean, limiting it to those with the two nominal affixes; he simply treats kes like other ‘defective nouns’ which lack a clear meaning. As for a functional category, the ‘lack of a clear meaning’ may not suggest an insignificant role; it would rather be its quintessential property. The categorial identity of kes as an N, however, remains intact in a construction like (18) despite the neutralization of its meaning. We will show that a clausal-nominalizer, kes is amore generalized one than the two nominal affixes -um and -ki; while selections of the affixes for in object NP are lexically controlled by a verb, kes is free from such a restriction. -Um and -ki are in the complementary distribution in the following sentences because of the different selectional features of the two verbs in the sentences of (19): (19) a. Suanei-ka [0i Suil-ul manna-ss-um(*ki)]-ul huwhoeha-ess-ta Nominal Functional Categories in Korean: ... Nom Acc 447 meet-Past-Comp-Acc regret-Past-Dec ‘Sunae regretted that she met Suil.’ b. Suil-i [Sunae-ka tola-o-ki(*m)]-lul Nom Nom kitaeha-ko-iss-0-ta return-come-Comp-Acc expect-ing-is-Pres-Dec ‘Suil is expecting that Sunae comes back.’ In contrast, noun kes can nominalize the constructions in the square brackets in the above sentences as the head of the noun-complement structure: (20) a. Sunaei-ka [0i Suil-ul manna-n ] Nom kes-ul hwuhoeha-ess-ta Acc meet-mod(past) thing-Acc regret-past-dec ‘Sunae regretted that she met Suil.’ b. Suil-i [Sunae-ka tolao-l ] Nom kes-ul kitaeha-ko-iss-0-ta Nom return-mod(fut) thing-Acc expect-ing-is-pres-dec ‘Suil is expecting that Sunae comes back.’ Therefore, despite Yoon H.-S.'s dismissal, noun kes in the noun-complement structure appears to be a more productive nominalizer than the two nominal affixes -um and -ki. In general, the head position of NP is assumed to be not a typical position for a functional element. Thus, one can ask the question how noun kes can occur in head position of the noun-complement construction as a clausal nominalizer. We assume that it is due to the two parametric properties of Korean, the head-final parameter and the phrase-final NP subordinator. The object NP in a sentence like (20a), in fact, has two stacked subordinators, the nominal subordinator of a clause, kes and the subordinator of NP, K (the Accusative maker) -ul. Its structure can be illustrated as in (21): 448 Mi-Jeung Jo (21) KP NP K S N -kes Nominal -ul Subordinator of NP Subordinator of a clause The structure (21) shows that the head-final position is converted into the subordinator of the preceding complement clause. Since noun kes is dominated by the N node its syntactic identity must be maintained. A semantically unspecific noun such as kes performs a grammatical role in the context formalized in (17); the semantically underspecified noun in the head position of an NP turns into a clausal nominalizer in the absence of an argument status in an complement clause. 3. Conclusion Since the definite determiner functions as the nominalization operator which shifts a property into a kind reference, its generic use appears to be a general property of languages with articles despite minor parametric variations (e.g. the nominalized adjectives in Greek versus adjectives modified by adverbials in Dutch and English). But such type-shifting operation is not applicable to all adjectives, and hence the definite determiner is a semantic nominalization operator. Nevertheless, in Modern Hebrew, according to Ritter(1991), the Nominal Functional Categories in Korean: ... 449 definite determiner is copied to an adjective modifying a noun.13 Even if this language has an article system, it seems implausible that the definite article has the type-shifting function turning an adjective into a kind reference. As for languages whose superstructure of NP is other than D, they do not seem to have a common sudordinator and nominalization operator. Korean happens to be a head-final language, and the subordinator of NP is the phrasal Case-marker which is adjacent to the head. We have shown that a specific noun such as kes can function as a syntactic nominalization operator for a clause in head position. It is due to its consistent syntactic properties as a noun and its semantic underspecification. It is an open question whether other articless language such as Chinese and Japanese also has a syntactic nominalization operator like Korean. Nevertheless, the existence of kes suggests that there can be at least one nominalization operator within an NP or a functional projection related to it. 13. According to Ritter's analysis of NPs in Modern Hebrew, a definite noun phrase and the construct state (CS) constructions such as beyt (ha)-mora (house (the)-teacher) ‘The teacher's house’ is DPs headed by D gen. However, modifying adjectives always agree in definiteness as well as number and gender, with the noun they modify. Definiteness agreement is indicated by the presence or absence of a copy of the definite article on the AP, as in the following examples (Ritter (1991:(7)): (1) a. yeladim nexmad-im ‘nice children’ children nice-Masc-Pl b. ha-yeladim ha-nexmad-im ‘the nice children’ the-children the-nice-Masc-Pl Given the copied definite article on an adjective in (1b), it is unlikely that D in Hebrew can function as a semantic nominalization operator as that it does in Greek and English. 450 Mi-Jeung Jo References Abney, Steven. 1987. The noun pnrase in its sentential aspect, unpublished Ph. D. dissertation, MIT Cheng, Lisa Lai-Shen and Rint Sybesma. 1999. Bare and not-so-bare nouns and the structure of NP, Linguistic Inquiry 30.4:509-542. Chierchia, Gennaro. 1998. Reference to kinds across languages, Natural Language Semantics 6:339-405. Fukui, Naoki. 1986. A Theory of Category Projection and Its Application, unpublished Ph.D. dissertation, MIT. Fukui, Naoki and Peggy Speas. 1986. Specifiers and projection, MIT Working Papers in Linguisticw 8:128-172. Giannakidou, Anastasia and Melita Stavrou. 1999. Nominalization and ellipsis in the Greek DP, The Linguistic Review 16.4:295-329. Giorgi, Alessandra and Giuseppe Longobardi. 1991. The Syntax of noun phrases: Configuration, parameters and empty categories, New York, NY.: Cambridge University Press. Jo, M.-J. 1999. The NP parameter and grammatical nouns in Korean, Korean Journal of Linguistics 23.3:447-475. Longobardi, Giuseppe. 1994. Reference and proper names: a theory N-movement in Syntax and Logical Form, Linguistic Inquiry 25.4: 609-665. Reuland, Eric. 1983. Govening -ing, Linguistic Inquiry 14:101-36. Ritter, Elizabeth. 1991. Two functional categories in noun phrases: nb evidence from modern Hebrew, In Syntax and semantics 25: Perspectives on phrase structure; heads and licensing, ed. Suan D. Rothstein, 37-62. San Diego, Calif.:Academic Press. Stowell, Tim. 1989. Subjects, Specifiers, and theX-Bar Theory, In Alternative conceptions of phrase structure, ed. by Mark R. Baltin and Anthony S. Nominal Functional Categories in Korean: ... 451 Kroch, 232-262. Chicago: University of Chicago Press. Szabolcsi, Anna. 1987 Functional Categories in Noun Phrase, In Approaches to Hungarian vol. 2 ed. by Istvan Kenesei, 167-190, Szeged, Hungary. Szabolcsi, Anna. 1994. The noun phrase, In Syntax and semantics 27: The syntactic structure of Hungarian, ed. Ferenc Kiefer and Katalin Kiss, 179-274. San Diego, Calif.:Academic Press. Yang, D.-W. 1972. Topicalization and relativization in Korean. Ph. D. nb disseration, Indiana University. Yim, Y.-J. 1984. Case-tropism: the nature of phrasal and clausal case. University of Washington doctoral dissertation. Yoon, H.-S. 1989. A Restrictive theory of morphosyntactic interaction and its consequences, University of Illinois at Urbana-Champaign dissertation.
© Copyright 2026 Paperzz