«Jeesun Nam Hankuk University of Foreign Studies Despite the numerous studies proposed to explain double nominative constructions (DNCs) in Korean, ...»
An Empirical Study of Korean Adjectival Language and Linguistics
Predicates that License Double Nominative © The Author(s) 2015
Reprints and permissions:
Hankuk University of Foreign Studies Despite the numerous studies proposed to explain double nominative constructions (DNCs) in Korean, empirical approaches to the predicate types that license their essential complements for conversion into nominatives have attracted little attention from scholars. However, given that, from syntactic and semantic viewpoints, the DNCs result from diverse and complex linguistic realities, a rigorous examination of the predicate types is required before attempting to elaborate theoretical assumptions further. Based on a large database of 6,600 Korean adjectives, this study distinguishes two types of DNCs: Predicate-Dependent DNCs (PD-DNCs), where the two nominatives are essential arguments of the adjectival predicates; and Discourse-Dependent DNCs (DD-DNCs), where one of the two nominatives is not related to the argument structures, but either is generated by the split of one argument or is newly introduced at the discourse level. About 30% of the 6,600 adjectives license PD-DNCs. On this basis, the current study provides two findings on PD-DNCs. First, adjectives licensing PD-DNCs are classified into five syntactic and semantic sub-types, the predominant complement type that is convertible into a nominative being that in -EY. Second, the adjectives licensing their complements for conversion into nominatives mostly show certain peculiar syntactic properties: most of them permit two syntactic constructions in a subject-complement crossing (SCC) relation like ‘N1-NOM N2-POS ADJ-DEC = N2-NOM N1-POS ADJ-DEC’. With regard to DD-DNCs, in this study the DNCs are divided into four sub-types, which exhibit various semantic and syntactic properties that are distinct from those of PD-DNCs.
Key words: adjective, complement type, conversion, discourse effect, double nominative
1. Introduction This article offers a new account of double nominative construction (DNC)1 phenomena in Korean, based on the examination of a large-scale database of adjectival predicates that occur in
the following sentence type:
* I am indebted to the anonymous reviewers and the editors of this volume for their valuable comments and suggestions on an earlier version of this paper. I would also like to thank Brendan Gillon for his insightful comments and suggestions. Thanks as well to Tianshu Liu, who helped me with the Chinese translation of the abstract. This work was supported by the Hankuk University of Foreign Studies Research Fund of 2015.
The terminology designating double nominative construction has undergone a change in Korean linguistics literature. At the beginning, it was often called DOUBLE SUBJECT CONSTRUCTION, which emphasized the aspect of the odd syntactic structure. As scholars started to doubt whether one really deals with two subjects, terminology such as DOUBLE NOMINATIVE CONSTRUCTION appeared. Still, the question about the syntactic nature of nominativemarked noun phrases has been raised, because semantic and syntactic relations among the noun phrases followed Jeesun Nam
N2-NOM ADJ-DEC2 (1) N1-NOM
In basic constructions, Korean adjectives occur as predicative elements without requiring a copula such as be in English. Since arguments are essentially accompanied by case-marking postpositions,
canonical sentence structures in Korean can be schematized roughly as follows:
(2) N1-NOM N2-POS N3-POS... Nn-POS (V+ADJ)-DEC Among diverse syntactic constructions representable by (2), (1) raises a particular issue, since this type of sentence is observed uncommonly frequently, and in addition, the adjectival predicates are extremely heterogeneous. The following examples, which each consist of two nominatives and an
adjectival predicate, illustrate this complex situation:
by the same type of case markers are heterogeneous in diverse examples. In this regard, the more prudent term, DOUBLE N-KA CONSTRUCTION (i.e. KA stands for the lexical form of nominative), has been suggested, in order not to assign a pre-acquired syntactic notion to these repeated noun phrases. However, ‘double N-KA’ is not a widely used term in Korean linguistics in comparison with ‘double nominative’. Consequently, the latter is adopted in the current study.
Abbreviations used in this paper are: ACC: Accusative; ADJ: Adjective; ADV: Adverb; CLA: Classifier;
CON: Conjunctive; DEC: Declarative; E: Empty Sequence; GEN: Genitive; HON: Honorification; KI:
Nominal Suffix -KI, roughly equivalent to the nominalization suffix such as -ing (e.g. solve → solv-ing);
LOC: Locative; NOM: Nominative (i.e. -ka or -i); NUM: Numeral; N1: First occurring Noun phrase; N2:
Second occurring Noun phrase; N3: Third occurring Noun phrase; Nn: Nth occurring Noun phrase; N-EY:
Complement in -EY (i.e. -ey or eykey, roughly equivalent to the preposition to, at, in, or for); N-KA: Complement in -KA (i.e. -ka or -i, roughly equivalent to the preposition in or at); N-WA: Complement in -WA (i.e. -wa or kwa, roughly equivalent to the preposition with or to); PAS: Past tense; POS: Postposition; PRE: Present tense; PRED: Predicate; SFX: Suffix; V: Verb; Vn: Nominalization of a verb.
Language and Linguistics 16(3)
As seen here, the syntactic and semantic functions of the nominatives in DNCs are not identical to one another. This observation makes persuasive the widely held assumption that this diversity results from the fact that not all DNCs are canonical structures for adjectival predicates. I agree with this commonly accepted assertion. The essential problem, however, is to know HOW to distinguish the canonical structures from the non-canonical ones, and WHAT is the most crucial factor for this distinction. Nevertheless, most previous works have converged on the explanation of WHY DNCs are different from each other on the basis of their own theoretical backgrounds. As a result, a limited number of selected examples have been examined to support the presupposed hypotheses.
I claim that, before attempting to explain why these complex DNC phenomena occur, all adjectival predicates licensing a DNC should be empirically examined. I examined in this study a database of 6,600 adjectival constructions proposed in Nam (1996, 2007), to obtain a set of adjectives licensing their complements for conversion into nominatives. I analyzed what types of complements required by adjectival predicates are sensitive to the conversion into nominatives. In this process, I distinguished two types of DNCs in Korean: the DNCs where the two nominatives are essential arguments of the adjectival predicates (named PREDICATE-DEPENDENT DNCS (PD-DNCS)), and the DNCs where one of the two nominatives is not related to the argument structure but either is generated by the split of one argument or is newly introduced at the discourse level (named DISCOURSEDEPENDENT DNCS (DD-DNCS)). Once the PD-DNC types are determined, the DD-DNCs are discussed on the basis of the diverse relations between two nominatives. Among the above examples, the first seven sentences (i.e. (3a)–(3g)) illustrate various PD-DNCs, whereas the second nine sentences (i.e.
(3h)–(3p)) show DD-DNC types. In the following sections, I shall discuss each of these DNC types.
Notice that DNCs, syntactically uncomfortable, have been considered one of the most problematic issues among Korean scholars (Choi 1937; Ko 1999; Lee 2002; D.-H. Lim 1997; H.-B.
Lim 1974; Nam 1986; Park 2001; Seo 1996; Song 2009, among others). The most vigorous discussion on DNCs has focused on determining the true subject from the two identical forms in DNCs.
In this regard, logical relation types between two nominatives are analyzed from semantic and discourse viewpoints and justified by syntactic criteria based on a theoretical foundation. Two assumptions are worthy of note. The ONE-SUBJECT ASSUMPTION, supported by M.-S. Kim (1971), Y.-H. Kim (1978), Lim (1997), Ko (1999), Ahn (2001), Park (2001), and Lee (2002), among others, postulates the existence of only one subject in DNCs, irrespective of the DNC type, and therefore, the other nominative form must be labeled in some way such as a newly added topic or a converted complement. Hence, one of the predominant concerns of the scholars has been to determine which one of them is the subject. The second assumption, the TWO-SUBJECT ASSUMPTION, held by Lim (1974), Choi (2002), K.-H. Kim (2003), Mok (2005), and Yoon (2007), among others, postulates that the two nominatives in DNCs are both subjects of the sentence, even though the nature of these two subjects or the scope of the predicates may be assumed differently according to their theoretical background. However, since there are diverse types of DNCs in Korean, building a unique theoretical hypothesis is not recommended. Moreover, DNCs are observed with adjectival predicates more frequently than with verbal ones. Thus, some scholars such as Ko (2001) or Song (2009) insisted that it would be indispensable to examine predicate types, notably adjectives, which license DNCs. None the less, no relevant studies on a sizeable number of adjectives have been proposed and only a few stereotypical adjectives have been considered in relation to certain problematic examples of double nominatives. In this study, based on the database of 6,600 adjectives, I advocate Language and Linguistics 16(3) two distinct DNC types which should be approached in a different manner regarding canonical structure issues such as the true subject determination. As a matter of fact, before I attempt to elaborate theoretical assumptions, a rigorous examination of the predicate types licensing PD-DNCs is required.
This article is organized as follows. Section 2 discusses previous works on DNCs in the Korean literature; §3 introduces five sub-types of the adjectival predicates licensing PD-DNCs, whereas §4 deals with four sub-types of DD-DNCs; and §5 presents the implications of the current study based on these empirical results.
2. Previous works on DNCs
The most active issues regarding DNCs in Korean have converged on the problem of recognizing the TRUE subject between the two nominatives. More precisely, they can be summarized as determining diverse logical relations between two nominatives from semantic and discourse viewpoints, and using a certain theoretical foundation to identify and justify the syntactic properties of the two nominatives.
Attempts to generalize semantic natures have often been blended with discourse notions such as the topic or focus of the utterance (Choi 2002; Doo 2010; Lim 1974; Park 2001). Semantic relations between two nominatives have been vigorously discussed for determining the subject of the sentence. According to Choi (2002), when the first occurrence of two nominatives in a CLASSMEMBER relation does not show semantic dependency with the predicate, it only emphasizes a ‘class’ that a ‘member’ occurring in the second noun position belongs to; thus, the second noun should be analyzed as the subject and the first one as the topic (e.g. Kyeycel-un kaul-i coh-ta ‘(As for) season, autumn is good = Autumn is the best season’). The most frequently discussed semantic relation types are related to the genitive phrases as in N1-uy N2 ‘N2 of N1’, which displays various logical relations between N1 and N2. For instance, in Choi (2002), POSSESSION and PART–WHOLE relations have been discussed, both of which are described as a BIG–SMALL relation in Song (2009) (e.g.
Chelswu-ka cha-ka mesci-ta ‘(As for) Chulsoo, the car is stylish=Chulsoo’s car is stylish’, and Khokkili-nun kho-ka kil-ta ‘(As for) elephants, the trunk is long=The trunk of elephants is long’).
In this regard, when two nominatives are analyzed as in this BIG–SMALL relation, the identification of the subject becomes a more complex problem. Choi (2002) and Mok (2005) have proposed N1 as the subject, whereas Ahn (2001) has claimed that N2 should be the subject. In contrast, Choi (2008) has proposed an analysis of N1 and N2 as double subjects.
In fact, regarding the identification of the subject, the previous works can be divided into two distinct approaches: one is based on the assumption that there is only one subject in DNCs, the non-subject being a converted complement or a topic word obtained by diverse channels (i.e. the One-Subject Assumption); and the other holds the assumption that there are two subjects in DNCs (i.e. the Two-Subject Assumption). When the latter assumption is supported, two viewpoints can be distinguished: some scholars claim that, in Korean, there are certain constructions composed of two syntactic subjects, while others assume that the first nominative is the subject of the so-called PREDICATIVE CLAUSE, which is, in turn, made of an inner subject and a predicate. The first viewpoint is the DOUBLE SUBJECT ASSUMPTION, supported by M.-S. Kim (1971), Y.-H. Kim (1978), Lee (2002), Jeesun Nam and Yoon (2007). The second one is the PREDICATIVE CLAUSE ASSUMPTION, held by Lim (1997), Ko (1999), Ahn (2001), and Park (2001). Here, the main issue is justifying their assumptions by using certain syntactic tests.
Among the prime syntactic evidence used to support these assumptions, the first piece of evidence is related to the presence of the HONORIFICATION SUFFIX -SI, which is known as being in
agreement with the subject. Let us compare the following sentences:
The presence of the honorification marker -SI in (4a) reveals that the second noun cungcopwu ‘great-grand-father’ is the subject, not the first one ku sonyen ‘that boy’. For the same reason, in (4b), this honorification marker cannot be attached, because the second noun aywankyen ‘pet dog’ is semantically inappropriate for being assigned with a honorification marker. Agreement with the honorification marker may be a criterion to determine the subject, and according to this criterion, N2 turns out to be the subject for these DNCs, not N1.