인문논총‧제51집‧2020.02. ISSN 2005-6222 / eISSN 2713-7511 pp.39~62 A Resultative Adverbial Approach to Korean Resultative Constructions JONGIL KWON Professor, Department of English, Kyungnam University; [email protected]Abstract: This paper presents a resultative adverbial approach to Korean resultative constructions (i.e., the [AP-key] constructions) against the null subject analysis in which secondary resultative predicate is a small clause containing PRO. On the basis of the observation that they are morpho-syntactically and semantically distinguished from English resultatives, I propose that some of resultative constructions in Korean can be analyzed as a resultative adverb. The most remarkable semantic feature of a resultative adverb is its capability to employ a local semantic subject. That is, a resultative adverb can have an in- dividual subject separated from the main verb. I argue in this paper that Korean resultative con- structions as resultative adverbs must not include any aspectual markers (e.g., –eci). Also, I argue that Korean resultative adverbs can be treated as a Pseudo resultative. After all, Korean resultative con- structions cannot be analyzed as a uniform semantic phenomenon. Keywords: resultative construction, secondary predicate, resultative adverb, small clause, null subject, implicit argument I. Introduction It has been commonly assumed that a resultative sentence has at least two predicates: the first and secondary predicates (Simpson, 1983; Kim, 1999; Wechsler & Noh, 2001; Shibagaki, 2013, and many others). In a resultative construction (henceforth, RC), secon- dary predicate indicates a certain state of result caused by the first predicate (i.e., a main verb). From a cross-linguistic perspective, secondary predicate of an RC is typically ex- pressed as a bare adjective phrase, as shown in (1) (hereafter, secondary predicate in bold): (1) Transitive Resultative Construction (English) a. John hammered the metal flat . b. John wiped the desk clean .
24
Embed
A Resultative Adverbial Approach to Korean Resultative … · 2020-02-28 · A Resultative Adverbial Approach to Korean Resultative Constructions45 In (6), the PredP (i.e., [AP-key])
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
A Resultative Adverbial Approach to Korean Resultative
Constructions
JONGIL KWON
Professor, Department of English, Kyungnam University; [email protected]
Abstract: This paper presents a resultative adverbial approach to Korean resultative constructions (i.e., the [AP-key] constructions) against the null subject analysis in which secondary resultative predicate is a small clause containing PRO. On the basis of the observation that they are morpho-syntactically and semantically distinguished from English resultatives, I propose that some of resultative constructions in Korean can be analyzed as a resultative adverb. The most remarkable semantic feature of a resultative adverb is its capability to employ a local semantic subject. That is, a resultative adverb can have an in-dividual subject separated from the main verb. I argue in this paper that Korean resultative con-structions as resultative adverbs must not include any aspectual markers (e.g., –eci). Also, I argue that Korean resultative adverbs can be treated as a Pseudo resultative. After all, Korean resultative con-structions cannot be analyzed as a uniform semantic phenomenon.
a. Chelswu-ka kumsok-ul napccakha-key twutulki-ess-ta.
C-NOM metal-ACC flat-KEY pound-PST-DEC
‘Chelswu hammered the metal so that it became flat.’
b. Chelswu-ka chayksang-ul kkaykkusha-key takk-ass-ta.
C-NOM desk-ACC clean-KEY wipe-PST-DEC
‘Chelswu wiped the desk so that it became clean.’
The RCs in (1) and (2) are generally classified as a transitive RC since they have a tran-
sitive main verb. Note that compared with English ones, Korean RCs in (2) typically in-
clude the suffix –key.2
The syntactic status of secondary predicate (i.e., the [AP-key]) appearing in Korean
RCs has been a vicious conundrum in the literature. Wechsler & Noh (2001) and Ko
(2015), for instance, claim that Korean secondary predicate is on a par with English coun-
terpart, and thus can be treated as a small clause. In this approach, forming a syntactic
unit together with the direct object of a main verb, secondary predicate is regarded as un-
saturated (or object-controlled). That is to say, it does not carry any independent argu-
ment within its syntactic domain. Instead, the direct object selected by a main verb plays
1. In addition to the [AP-key] construction, Korean shows a different type of RC; namely, the postpositional phrase [DP-lo], as shown in (i):
(i) Chelswu-ka okswuswu-lul kalu-lo ppah-ass-ta. Chelswu-NOM corn-ACC powder-LO pound-PST-DEC ‘Chelswu pounded the corns into power.’ In (i), the resultative phrase kalu-lo ‘into powder’ predicates of a changed state resulted from the action of the
main verb. However, this paper does not take this type of RC into consideration. 2. The exact morpho-syntactic nature of the marker –key is still controversial in the literature. Kim (1999), in
particular, treated it as a (secondary) predicator whereas Wechsler & Noh (2001) analyzed it as just a complementizer. Recently, Ko (2015) notated it as a relator by adopting den Dikken (2006)'s view that small clause is uniformly expressed as a bare adjective phrase containing a null subject (i.e., PRO). In general, along with the suffix -(h)i, the suffix –key is regarded as an adverbial marker, which is morphologically attached into an adjectival stem (e.g., alumdap-key 'beautifully,' noph-key 'highly,' pwutulep-key 'smoothly,' etc.). In this paper, I will notate it as –KEY.
41A Resultative Adverbial Approach to Korean Resultative Constructions
a logical subject for it (Li, 1999; Wechsler & Noh, 2001; Song, 2005; Yeo, 2006, inter
alia).
On the other hand, Shim & den Dikken (2007) and Lee (2016) claim that the Korean
RCs in (2) can be treated as an adjunct. In pursuing a unified theory of Korean RCs, they
claim that unlike English ones, Korean RCs are fully saturated, and hence all the secon-
dary predicates in (2) include an implicit local subject (i.e., pro) in their clausal domain
(i.e., TP). To put it differently, Korean does not adopt a small clause-based structure for
the RC; rather, it is an adjunct clause carrying pro in its local argument position.
In this paper, I assume that the RCs in (3) are radically distinct from the ones in (2) in
certain respects:
(3) a. Chelswu-ka kumsok-ul/-i napccakha-eci-key
C-NOM metal-ACC/-NOM flat-ASP-KEY
twutulki-ess-ta.
pound-PST-DEC
‘Chelswu hammered the metal so that it became flat.’
b. Chelswu-ka chayksang-ul/-i kkaykkusha-eci-key
C-NOM desk-ACC/-NOM clean-ASP-KEY
takk-ass-ta.
wipe-PST-DEC
‘Chelswu wiped the desk so that it became clean.’
Unlike the latter, the former includes the aspectual marker –eci and is very sensitive to
certain semantic restrictions that are commonly observed in English RCs. Based on the
difference between (2) and (3), I propose that only the RC in (3) is a true type of RC. The
presence of an aspectual marker –eci seems to be crucial to determine whether the RC is a
true type or not. In contrast, I will treat the RC in (2) as a pseudo type of RC. Especially,
I will argue that the secondary predicate in (2) can be classified as a resultative adverb in
42 Jongil Kwon
the sense that it includes an independent local subject.3
Against the major view that Korean RC is a small clause containing a null subject, in
this paper, I will provide a resultative adverbial analysis on the RC type given in (2).
Meanwhile, I focus on three grammatical issues: (i) how Korean RCs are distinct from
English ones, (ii) how they take their semantic subject, and (iii) whether or not they are al-
ways analyzed as a resultative adverb.
This paper is organized as follows. Chapter 2 deals with a variety of differences be-
tween English and Korean RCs. It will be shown that the existing analyses are not proper
to capture the diverse syntactic or semantic properties of Korean RCs. Chapter 3 is my
proposal. It will be argued that some of Korean RCs (but not all) can be analyzed as a re-
sultative adverb, and that a resultative adverbial analysis stands on the basis of empirical
evidence. Chapter 4 is conclusion.
II. English vs. Korean Resultative Constructions
An RC usually includes a secondary predicate which describes a changed state resulted
from the action triggered by the first predicate (i.e., a main verb). In the literature, de-
pending on the syntactic relation between secondary predicate and its logical subject, RCs
has been classified into three subtypes: transitive, unaccusative, and unergative RCs
(Simpson, 1983; Kim & Maling, 1997; Kim, 1999, and many others). In general, English
RCs are subject to the Direct Object Restriction (DOR) effect, whereby the logical subject
of an RC must be identical with the direct object selected by a main verb. Among the
three types of RCs, a transitive RC always occurs in the sentence including the direct ob-
ject subcategorized by a transitive main verb. The transitive RC is often called an
‘unsaturated’ or ‘object-controlled’ resultative since the direct object is associated with the
null subject (i.e., PRO) of small clause, as already mentioned in the previous chapter.
3. A resultative adverb refers to an unusual type of adverb which can take an implicit argument individually cre-ated by the event of a main verb (Geuder, 2000; Levinson, 2010).
43A Resultative Adverbial Approach to Korean Resultative Constructions
On the other hand, the other two types of RCs are all connected with intransitive main
verbs: unaccusative and unergative verbs. The unaccusative RC, which is often called a
‘subject-controlled’ RC, is derived by the unaccusative intransitive verb, as shown in (4):
(4) Unaccusative Resultative Construction (English)
a. The pond froze solid.
b. The bread burned black.
(5) Unaccusative Resultative Construction (Korean)
a. Yenmos-i tantanha-key el-ess-ta.
pond-NOM solid-KEY freeze-PST-DEC
‘The pond froze so that it became solid.’
b. PPang-i kkamah-key ta-ss-ta.
bread-NOM black-KEY burn-PST-DEC
‘The bread burned so that it became black.’
In the unaccusative RC (4a), something (not appear in the actual sentence) causes the
pond to be solid by the result of freezing. The secondary predicate solid reflects a change
of state that results from the event denoted by the main verb freeze. At first glance, this
type of RC is not subject to the DOR that is critical to define English RCs. According to
Chomsky (1995), however, the subject of an unaccusative verbal construction is a final
syntactic outcome derived from its theta role (i.e. a direct object) position to accomplish
its morphological requirement (i.e. nominative case). In other words, the subject of the
unaccusative RC (4a) is syntactically derived from its direct object position, and thus the
matrix subject and the null subject of secondary predicate together form a semantic unit
in their base-generative positions. This argumentation strongly indicates that the un-
accusative RC as well as the transitive RC is sensitive to the DOR in the way that they
share the same structure in the earlier linearization of narrow syntax.
44 Jongil Kwon
The semantic relationship between direct object and secondary predicate in the tran-
sitive and unaccusative RCs has been an important tool to support the ‘small clause’ anal-
ysis, in which secondary predicate of RC is a bare AP carrying a null subject (Simpson,
1983; Carrier & Randall, 1992; Levin & Rappaport Hovav, 1995, inter alia). Following
Levin & Rappaport Hovav (1995), a majority of studies have incessantly suggested that
the syntactic structure of Korean RCs is substantially parallel to the English counterpart
2009; Ko, 2015, and many others). Under this null subject approach, an RC is a mini-
mum clause in the sense that it consists of a null subject and a predicate, but cannot carry
any verbal inflections, modifiers, or internal arguments that are usually found in the fully
projected clause (i.e., TP or CP). The ‘unsaturated’ relationship between secondary predi-
cate and its logical subject is syntactically realized as PRO, as roughly illustrated in (6) (cf.
Ko, 2015, p.357):4
(6) A Small Clause Analysis of (2a)
... VP
DP V’
chayksangi-ul ‘desk-ACC’ PredP (=SC) V
PROi Pred’
AP Pred
kkaykkusha ‘clean’ -key
4. Since the categorial nature of –key, which introduces a resultative predicate, is still controversial among Korean linguists, I will gloss it just as ‘Pred’ in the sense of Kim (1999).
45A Resultative Adverbial Approach to Korean Resultative Constructions
In (6), the PredP (i.e., [AP-key]) is a small clause including PRO and kkaykkusha
‘clean.’
Another noticeable feature with regard to the structure (6) is about the syntactic rela-
tionship between the main verb and the secondary predicate. In (6), the PredP is a com-
plement of the main verb, not an adjunct. The dispute on whether it is a complement or
an adjunct, in fact, has been another issue among Korean linguists. Kim & Maling (1997),
Kim (1999), and Lee (2016) claim that Korean resultatives can be uniformly treated as
complementation. Song (2005), Yeo (2006), and Ko (2015) are stick to a hybrid analysis
in which some of Korean RCs can be alternatively analyzed as an adjunct. Interestingly,
Shim & den Dikken (2007) attempts to argue that Korean RCs can be uniformly charac-
terized as an adjunct. The controversy over complementation vs. adjunction is heavily re-
lied on whether or not to accept the basic assumption that Korean RCs are on a par with
English ones syntactically and semantically.
The unergative RC has fueled an argument over the small clause analysis. It is note-
worthy that the English RC in (7) introduces the ‘fake’ direct object, which is not sub-
categorized by the main verb. In other words, despite the fact that there is no direct the-
matic connection between the unergative main verb and its direct object, the unergative
RC requires the ‘fake object’ (e.g. himself or his shoes) in order to satisfy the DOR. It is
clear that a fake object plays a controller role of the following resultative predicate
(Simpson, 1983; Levin & Rappaport Hovav, 1995). This is exactly why the unergative RC
is often called an ‘unselected’ resultative in the literature (Wechsler & Noh, 2001; Son,
2008; Lee 2009, and many others.).
It is noticeable that there is a significant discrepancy between English and Korean un-
ergative RCs. While English one requires a fake object to satisfy a legitimate structure
(i.e., the DOR), Korean unergative RC is totally free from such morpho-syntactic
constraint. In Korean, secondary predicate of the unergative RC is regarded as a fully
‘saturated’ clause in that it carries a non-null subject in its external argument position, as
exemplified in (8):
46 Jongil Kwon
(7) Unergative Resultative Construction (English)
a. John cried *(himself) hoarse.
b. John ran *(his shoes) threadbare.
(8) Unergative Resultative Construction (Korean)
a. Chelswu-ka mok-i/*-ul swi-key solichi-ess-ta.
C-NOM throat-NOM/-ACC hoarse-KEY shout-PST-DEC
‘Chelswu shouted so (loud) that his throat became hoarse.’
b. Chelswu-ka sinpal-i/*-ul talh-key talii-ess-ta.
C-NOM shoes-NOM/-ACC threadbare-KEY run-PST-DEC
‘Chelswu ran away so that his shoes became threadbare.’
The availability of the nominative case-marked subject in (8) strongly indicates that
some of Korean resultatives can be radically different from English resultatives. Wechsler
& Noh (2001), in particular, argues that unlike English one, Korean unergative RC does
not hire the ECM mechanism since the logical subject of secondary predicate is directly
assigned both case and a thematic role by the secondary predicate. In this regard, adopt-
ing a split analysis of RCs, Lee (2009) suggests that in Korean, the unergative RC needs to
be separately treated from the other types of RCs (i.e., the transitive and unaccusative
RCs).
On the other hand, Shim & den Dikken (2007) claims that Korean RC always takes a
‘fully saturated’ secondary predicate. Against the major view that Korean RC is a small
clause-based complement, they argue that in Korean, every RC has an adjunct structure
projected to a TP in a uniform way. Under their adjunction-based analysis, secondary
predicate is an adjunct clause, which is adjoined to vP (or VP) in the matrix clause. For ex-
ample, the secondary predicate in (8a) includes pro co-indexed with the matrix subject.
Note that the pro forms a constituent (i.e., DP) together with the subject DP mok ‘throat’
in the so-called ‘inalienable (possessor-possessee)’ relation, as roughly illustrated in (9) (cf.
Shim & den Dikken, 2007, p.7):
47A Resultative Adverbial Approach to Korean Resultative Constructions
(9) An Adjunct Analysis of (8a)
... vP
TP (resultative) vP
[DP proi mok]-i swi-key DP v’
‘throat-NOM hoarse’ Chelswui–ka ‘C-NOM’ VP v
solichi- ‘shout’
In (9), the nominative case appearing in the RC is an outcome of TP. Also, the pro is a
possessor of the local subject mok-i ‘throat-NOM.’ This structure is very different from
what we have seen in (6) postulating the null subject (i.e., PRO).
The across-the-board style of analysis suggested by Shim & den Dikken (2007), how-
ever, cannot capture all the diverse syntactic and semantic properties of Korean RCs.
Above all, given the assumption that an RC is a TP adjunct, it is quite predicable that sec-
ondary predicate can accompany with a tense marker such as the present or past verbal in-
flection (e.g., -(nu)n or –a/ess). Unfortunately, this is not borne out, as demonstrated in
‘Chelswu hammered the metal in two hours so that it
became flat.’ (telic only)
50 Jongil Kwon
In terms of the telicity effect, the Korean RC in (14) containing the inchoative marker
–e/aci are exactly in line with the English one in (11b), which does not allow any in-
choative verbal form. Although there is some chiastic morphological relationship between
English and Korean RCs, the telicity restriction leads us to assume that the RC in (14) is
a true type of RC in Korean, but not the one in (12).
Another semantic difference between English and Korean RCs provides a strong evi-
dence in favor of the assumption that the RC in (14) is substantially the counterpart of the
English RC (11b). It has been often suggested in the literature that there is a strong se-
mantic connection between the first predicate (i.e., a main verb) and the secondary predi-
cate in the RC (Washio, 1997; Lee, 2016, inter alia). In English, for example, the main
verb wipe in (15b) directly entails the result denoted by the secondary predicate clean.
This semantic accordance between two predicates is also captured in (16b). However,
(16a) is relatively free from such semantic restriction, and thus it fits with the un-
predictable result denoted by the adjective telepta ‘dirty’:
(15) a. John hammered the metal flat/(*thick)
b. John wiped the desk clean/(*dirty)
(16) a. Chelswu-ka chayksang-ul kkaykkusha-key/telep-key
C-NOM desk-ACC clean-KEY/dirty-KEY
takk-ass-ta.
wipe-PST-DEC
‘(Lit) Chelswu wiped the desk clean/badly.’
b. *Chelswu-ka chayksang-ul telep(w)-eci-key
C-NOM desk-ACC dirty-ASP-KEY
takk-ass-ta.
wipe-PST-DEC
‘Chelswu wiped the desk so that it became clean.’
51A Resultative Adverbial Approach to Korean Resultative Constructions
This implies that only the RC including the aspectual marker –eci is regarded as a true
type of RC with regard to semantic restrictions.
In sum, we have discussed a couple of semantic properties in Korean RC, which is
sharply distinct from English one, A semantic approach to Korean RCs suggests that the
RC in (14) containing the inchoative marker –eci is a true counterpart of the English RC.
Meanwhile, the Korean RC in (12), which has been hitherto regarded as a null subject RC,
appears to be a pseudo type of resultative. In the next section, we will explore the nature
of a pseudo type of RC in detail.
2. Proposal: A Resultative Adverbial Analysis
In the previous section, we have seen that only the RC in (14) carrying the inchoative
suffix -eci is parallel to the English RC with respect to a couple of semantic agreement. In
this section, it will be argued that the [AP-key] construction in (12) is a resultative adverb
rather than a small clause. Also, this section provides a syntactic and semantic account for
how a simple adverb can take its semantic subject like a resultative predicate.
Geuder (2000) observed that a special type of English adverb describes something cre-
ated by the event denoted by a main verb, as given in (17) (cf. Geuder, 2000, pp.81-84):
(17) a. John decorated the room beautifully/(*beautiful).
(i) So, the decoration of the room is beautiful.
⇒ Resultant argument: beautiful decoration
(ii) #So, the room is beautiful.
b. John loaded the cart heavily/(*heavy).
(i) So, the load of the cart is heavy.
⇒ Resultant argument: heavy load
(ii) #So, the cart is heavy.
Unlike a manner adverb (e.g., slowly or quickly) describing how the action is con-
52 Jongil Kwon
ducted, the adverbs in (17) are very similar to the secondary predicate of English RC in
that they denote the changed state caused by the event of a main verb. The adverb beauti-
fully in (17a), so to say, does not describe the action of decorating, but the result state of
the decoration event. Geuder called this a ‘resultative adverb’ in the sense that it describes
something lexically hidden in the event of a main verb. Unlike secondary resultative pred-
icate whose semantic subject is always associated with the direct object (i.e., the DOR),
the resultative adverbs in (17) have their own implicit subjects that are contextually de-
termined by the semantics of a main verb. According to Geuder, the implicit subject can
be easily recovered by the so-called ‘result nominalization’ such as beautiful decoration or
heavy load, as demonstrated in (17). This explains exactly why in (17), the interpretation
(i) is much better than the one (ii).
Similarly, Levinson (2010) showed that although they are not semantically distinct
from the typical resultative adjectives, the adjectives in (18) do not semantically modify
the direct object. For instance, Mary’s hair does not become tight as a result of her braid-
ing; rather, what becomes tight is her braid implicitly created by braiding. Note that the
implicit argument of adjective (e.g., braid and pieces) is not overtly realized in the syntax,
but is pragmatically created from the event of a main verb, as shown in (18) (cf. Levinson,
2010, P.137):
(18) a. Mary braided her hair tight.
⇒ Her braid is tight. vs. #Her hair is tight.
b. Mary chopped the parsley fine.
⇒ The chopped pieces are fine. vs. #The parsley is fine.
Levinson claims that the adjectival construction in (18) is a ‘pseudo’ resultative in that
it modifies a covert argument contextually (or pragmatically) derived by the lexical en-
tailment of a main verb. The pseudo RC is not syntactically and semantically parallel with
the true RC, which always modifies the direct object (i.e., the DOR).5
53A Resultative Adverbial Approach to Korean Resultative Constructions
Nakazawa (2012) argues that the semantic subject of an RC is not always identical
with the argument selected by the unaccusative verb in Japanese and Korean, as illus-
trated in (19) and (20) (cf. Nakazawa, 2012, p.598):
(19) a. Kawa-ga atu-ku koot-ta
river-NOM think-KU froze-DEC
‘(Lit) The river froze thick.’
b. Kang-i twukkep-key el-ess-ta
river-NOM thick-KEY freeze-PST-DEC
‘(Lit) The river froze thick.’
(20) a.?*Kangi-i [PROi twukkep-key] el-ess-ta.
river-NOM thick-KEY freeze-PST-DEC
‘(Int) #The river froze, so the river is thick.’
b. Kang-i [(elum-i) twukkep-key] el-ess-ta.
river-NOM ice-NOM thick-KEY freeze-PST-DEC
‘(Int) The river froze, so (the ice) became thick.’
In (19) and (20), the [AP-ku/-key] construction in Japanese and Korean is clearly dis-
tinct from the typical RC that we have seen in the previous chapter. Particularly, it does
not directly modify the subject of the unaccusative verb; rather, it modifies a contextually
created argument (i.e., ice), which is not phonetically realized in the sentence whatsoever.
What is noteworthy in (19) and (20) is that the previous analysis based on the null subject
structure cannot capture the semantic disagreement expressed in (20b).6 Furthermore,
5. Levinson (2010) claims that the pseudo RC is not semantically identical with the resultative adverb suggested in Geuder (2000). The semantic criteria to distinguish them suggested by her are more or less unclear. In this paper, I will treat both resultative adverb and pseudo resultative as the same category.
6. Wechsler & Noh (2001, p.409) claims that the Korean word twukep-key ‘thick-KEY’ is not a resultative, but a manner adverb (i.e., ‘The river froze in a thick manner.’). This reading is acceptable, albeit marginally. But their analysis cannot still capture the reading in (20b).
54 Jongil Kwon
the [AP-key] twukkep-key ‘thick-key’ in (20b) cannot be analyzed as a manner adverb
since it can occur with its local subject (cf. Wechsler & Noh, 2001). According to Geuder
(2000) and Levinson (2010)’s analysis, the local argument (e.g., the ice in (20b)) is a sine
qua non in the resultative adverbial construction. Note that although it is semantically
created as an entailment of a main verb, the local subject can vary depending on the prag-
matic environments where the event happens.
Adopting their insightful findings, I propose that some of Korean RCs (i.e., the
[AP-key] construction in (2) and (5)) can be analyzed as a resultative adverb, but not a
true type of RC. This resultative adverb is unusual in that it can introduce its own subject
to form a resultative predicate (i.e. a fully saturated clause). To put it differently, against
the small clause-based analysis in which secondary predicate always carry a null subject, I
suggest that the [AP-key] construction is a resultative adverb, and that it is a fully satu-
rated phrase containing an implicit subject. Moreover, the local subject can be realized as
an explicit form, as seen in (20b). The following structure (21) represents a resultative ad-
verb construction for the unaccusative RC in (19b) and (20b):
(21) An Resultative Adverb Analysis of (19b) and (20b)
... VP
DP V’
kang-i ‘river-NOM’ AdvP (resultative adverb) V’
(ei or elumi-i ‘ice-NOM’) Adv’ V
AP Adv el- ‘freeze’
twukkep ‘thick’ -key
55A Resultative Adverbial Approach to Korean Resultative Constructions
Note that the resultative adverb in (21) is an adjunct of the main verb, and has its in-
dividual subject, independent from the matrix arguments (i.e., direct object and subject)
selected by the main verb. I assume that the subject of a resultative adverb can be either
implicitly or explicitly realized, and is assigned nominative case as a default case (cf. Jang
& Kim, 2001).
3. Consequences
There is some empirical evidence supporting the resultative adverb analysis of Korean
[AP-key] constructions. First of all, depending on various pragmatic considerations, di-
verse local subjects can be showed up in the resultative adverb constructions, as exempli-
fied in (22):
(22) a. Chelswu-ka kumsok-ul (kuth-i/moseli-ka)
C-NOM metal-ACC end-NOM/corner-NOM
napccakha-key twutulki-ess-ta.
flat-KEY pound-PST-DEC
‘Chelswu hammered the metal so that its end/its corners
became flat.’
b. Chelswu-ka chayksang-ul (pyomyeon-i/kacangcali-ka)
C-NOM desk-ACC surface-NOM/edge-NOM
kkaykkusha-key takk-ass-ta.
clean-KEY wipe-PST-DEC
‘Chelswu wiped the desk so that its surface/its edge became clean.’
The local subject of the [AP-key] in (22) is totally unpredictable under the null subject
hypothesis demanding a small clause carrying PRO. However, under the resultative ad-
verb analysis, the implicit or explicit local subject of the resultative adverb can be man-
ifoldly determined by the pragmatic information created by the event of a main verb. The
examples in (22) also show that the local subject of a resultative adverb enters into the
56 Jongil Kwon
possessor-possessee relation with the controlled argument of the matrix clause.
Second, a coordination structure strongly supports that the [AP-key] in (23) is an
adverb. To be more concrete, given that coordination happens only when the two con-
joined constituents are identical with each other, the [AP-key] constructions in (23) are
conjoined with the typical manner adverbs such as ppalli ‘quickly’ or yelsimhi ‘diligently.’
Thus, they must be analyzed as an adverb but not an adjective:
(23) a. Kang-i [twukkep-key (kuliko) ppalli] el-ess-ta
river-NOM thick-KEY and quickly freeze-PST-DEC
‘(Lit) The river froze thick and quickly.’
b. Chelswu-ka chayksang-ul [kkaykkusha-key (kuliko)
C-NOM desk-ACC clean-KEY and
yelsimhi] takk-ass-ta.
diligently wipe-PST-DEC
‘(Lit) Chelswu wiped the desk clean and diligently.’
Third, as a resultative adverb, the [AP-key] in (24a) cannot be replaced by the typical
resultative adjunct clause –tolok ‘until.’ This implies that the [AP-key] construction in
(24a) is not a true type of RC. In contrast, note that the [AP-eci-key] in (24b) as a true
type of RC can be substituted for by the –tolok adjunct clause, as demonstrated in (24b):
(24) a. Chelswu-ka kumsok-ul (napccakha-key)/
C-NOM metal-ACC flat-KEY
(?*napccakha-tolok) twutulki-ess-ta.
flat-until pound-PST-DEC
‘Chelswu hammered the metal so that it became flat.’
b. Chelswu-ka kumsok-ul (napccakha-eci-key)/
C-NOM metal-ACC flat-ASP-KEY
57A Resultative Adverbial Approach to Korean Resultative Constructions
(napccakha-eci-tolok) twutulki-ess-ta.
flat-ASP-until pound-PST-DEC
‘Chelswu hammered the metal so that/until it became flat.’
Fourth, in Korean, the topic maker –(n)un or the delimiter –man ‘even/only’ is usually
attached into adverbs but not adjectives (Lee, 2016). In this regard, the [AP-key] con-
structions in (25) must be counted as an adverb, as given in (25):
(25) a. Kang-i [twukkep-key-nun/-man] el-ess-ta
river-NOM thick-KEY-TOP/-DEL freeze-PST-DEC
‘(Lit) The river froze thick.’
b. Chelswu-ka chayksang-ul [kkaykkusha-key-nun/-man]
C-NOM desk-ACC clean-KEY-TOP/-DEL
takk-ass-ta.
wipe-PST-DEC
‘(Lit) Chelswu wiped the desk clean.’
Fifth, a resultative adverb occurs in the syntactically designated position for adverbs,
as given in (26):
(26) a. [[kkaykkusha-key/kkaykkushi] dakk-un] chayksang
clean-KEY/cleanly wipe-ADN desk
‘the cleanly wiped desk’
b. [[tantanha-key/tantanhi] e(l-u)n] kang
solid-KEY/solidly freeze-ADN river
‘the solidly frozen river’
Note that the resultative adverbs in (26) can be freely alternated with the manner ad-
verbs (e.g., kkaykkushi ‘cleanly’ or tantanhi ‘solidly’).
58 Jongil Kwon
Sixth, scrambling provides empirical evidence for the assumption that the [AP-key]
construction is an adjunct, but not a complement. The [AP-key] construction in (27) is
subject to the so-called ‘adjunct island effect,’ whereby nothing can be extracted from the
adjunct domain. The local subject of the resultative adverb cannot undergo scrambling,
as demonstrated in (27b):
(27) a. Chelswu-ka chayksang-ul [ pyomyeon-i
C-NOM desk-ACC surface-NOM
kkaykkusha-key] takk-ass-ta.
clean-KEY wipe-PST-DEC
‘Chelswu wiped the desk so that its surface/its edge became clean.’