한국언어정보학회 소 식 제 9 호 (1997년 12월) KSLI Newsletter (Korean Society for Language and Information) No. 3, April 1997 우)449-791 경기도 용인시 모현면 왕산리 산 89 한국외국어대학교
인문대학 언어학과내 전화: 0335-30-4293/4286
전송: 0335-30-4074 e-mail: hrchae@maincc.hufs.ac.kr ◆ 네 번째 모임 안내: 12월
13일 (토), 9:30 a.m., 대우재단 빌딩 이번 학기 마지막 모임인 네 번째 모임은 12월 13일 (토) 오전 9시 30분부터 대우재단 빌딩에서 열립니다. 많이 참석해서 유종의 미를 거둘 수 있도록 해 주시기 바랍니다. ◆ 언어와 정보 표지 디자인
공모 우리 학회지 언어와 정보 창간호가 12월 중에 출판될
예정입니다. 이 책의 표지 디자인에 대한 좋은 의견이 있으시면 편집인에게
알려 주시든지 구체적으로 도안을 해서 12월 13일의 모임에 가져 오시면
고맙겠습니다. 시일이 촉박한 관계로 이 모임에서 표지의 모양을 결정해야만
할 것 같습니다. ◆ 원고 모집, 발표자 모집
언어와 정보 2권 1호가 1998년 6월에 발간될 예정입니다.
이 책에 실릴 원고를 모집하고 있으니 98년 3월 31일까지 편집인 이현우
교수 (인하대학교 영어교육과)에게 보내 주시기 바랍니다 (032-860-7857
(O), 032-816-6795 (H), hylee@dragon.inha.ac.kr). 원고 작성 방법에
대해서는 Linguistic Inquiry 24권 1호 (1993년 겨울호) 185-197쪽을
참조하시면 됩니다. 그리고 1998학년도 1학기 연구발표회에서 논문을
발표하고자 하시는 분이나 주변에서 추천할 만한 분이 있으면 언제든지
총무에게 알려 주시기 바랍니다. ◆ 12월 13일 모임 발표자의
논문 초록 Wh-interrogatives and a Theory of Quantification 유 은 정 (서울대학교) This paper investigates quantification phenomena and wh-interrogatives, and provides a unified account of scope for both quantificational noun phrases and wh-phrases. Within transformational grammar, it has been generally assumed that logical form is represented in a separate level of syntactic representation, and scope of quantificational NPs and wh-phrases are determined by movement of such phrases (viz. quantifier raising and wh-movement). In this paper, we propose a precise analysis of quantifier and operator scope within the framework of Head-Driven Phrase Structure Grammar (HPSG), wherein the set of well-formed structural representations, which simultaneously contain LF-like and SS-like representations as subparts, is defined by a system of recursive feature constraints. Our theory of quantifier scope is an extended and revised version of Pollard & Sag's (1994) theory that utilizes Cooper's (1983) quantifier storage mechanism. The proposed theory constitutes a significant advance over Pollard & Sag in that it provides a solution for one of the major problems with Pollard & Sag, viz. the account of narrow scope readings in raising verb constructions and unbounded dependency constructions, by "lexicalizing" the quantifier store mechanism. Our account of wh-questions is based on the revised quantifier scope theory.
We investigate syntactic properties of English wh-question, and show how such properties as pied-piping
play a role in determining interrogative scope. We propose that
wh-scope in a
syntactic wh-movement
language like English requires syntactic licensing in addition to
the usual quantifier storage mechanism. Our analysis of English
interrogatives accounts for the puzzling asymmetry between subject-wh-questions and nonsubject-wh-questions with respect
to possible interrogative scoping. Our analysis also provides an
account for interesting scope facts regarding amount wh-phrases (e.g. how many books). We extend
our approach to wh-in-situ
languages as well and show how interrogative scope is determined
in this type of languages. A Pragmatic Analysis of Quantificational Force in Wh-phrases 홍 민 표 (명지대학교) This paper proposes to analyze the universal quantificational
force found in Korean [wh- + -to]
constructions in terms of such notions as focus and alternatives,
illocutionary operators, and scalar implicatures. The Japanese equivalent
[wh- + -mo] is
analyzed, in Nishigauchi (1986, 1990), as a universal quantifier
whose quantificational force is largely attributed to the particle
'-mo'. He treats
'-mo' as an unselective
universal quantifier that binds the free variable introduced by
the wh-phrase since a wh-phrase is identified as an indefinite in
Kamp (1981) and Heim's (1982) sense. I point out that any attempts
to apply Nishigauchi's idea to Korean [wh-
+ -to] constructions will have to resort
to the lexical ambiguity of the particle '-to' in Korean, which shows diverse meanings as a simple
additive marker, as a conjunctive marker, and as a concessive marker.
Instead, I propose to account for the universal quantificational
force in terms of semantic strength and the speakers' intention
in uttering such a sentence, by defining the illocutionary operator
of Emphatic Assertion (Krifka 1994) and the felicity conditions
along with it. This approach will allow a uniform analysis of the
Korean additive particle '-to.' 서술명사의 정체성과 통사적 정의 기준 채 희 락 (한국외국어대학교)
'서술명사'는 명사적 성질과 동사적 성질을 동시에 가지고 있기 때문에 이에 대한 분석이 참으로 다양하다. 이들을 (명사가 아니라) 동사로 분석하기도 하지만 (H-D Ahn 1991, K Park 1995), 대부분의 학자들은 품사상으로는 명사라는 것에 동의를 한다. 명사로 분석을 하더라도 이들의 통사적인 역할이 어떻게 되느냐에 대한 의견이 분분하다. 서술명사가 하위범주화만 담당하고 있다는 입장 (J Yoon 1991)과 하위범주화뿐만 아니라 격부여의 역할도 담당하고 있다는 입장 (채희락 1996, H-R Chae 1996)으로 대별된다. 다른 한 편으로는 서술명사라는 것이 (보충어를 가질 수 있는) 일반 명사와 통사적으로는 아무런 차이가 없다는 주장을 펴는 학자도 있다 (노용균 1997). 이 논문에서는 우선 서술명사가 통사적으로 일반
명사와는 다른 성격을 보이고 있음을 밝힘으로써 하나의 독립된 명사
하위부류로 정립되어야 함을 보이려고 한다. 아울러 이들의 성질을 규명하기
위해서는 어떤 종류의 구조를 설정해야 할지에 대한 논의도 곁들일 것이다.
이러한 논의의 바탕 위에서 어떤 명사가 서술명사인지를 판별할 수 있는
통사적 기준에 대해서도 살펴 볼 것이다. <참고 문헌> 노용균 (1997) "한국어 동사와 명사 사이의 하위범주화에 있어서의 평행성", 언어와 정보 1. 채희락 (1996) "하-의 특성과 경술어 구문", 어학연구 32.4, 409-476. Ahn, Hee-Don (1991) Light Verbs, VP-movement, Negation and Clausal Architecture in Korean and English, Ph.D. dissertation, Univ. of Wisconsin-Madison. Chae, Hee-Rahk (1996) "Light Verb Constructions and Structural Ambiguity," PACLIC 11, 99-107. Park, Kabyong (1995) "Verbal Nouns and Do-Insertion," Studies in Generative Grammar 5.1, 319-358. Yoon, James H-S (1991) "Theta Operations and the Syntax of Multiple Complement Constructions in Korean," Harvard
Studies in Korean Linguistics 4, 433-445.
Numeral Classifiers and Quantization 이 정 민 (서울대학교) Those numeral classifier languages like Korean, Japanese, and Chinese mobilize classifiers together with numerals in naming and counting objects in the world. This paper is concerned with characterizing the numeral classifier construction in Korean in connection with quantization (Krifka 1997) effects. In English, indefiniteness is represented by the bare plural/mass noun, and the indefinite article is used to denote quantization with change of state verbs, etc. as follows: (1) Mary ate apples/drank wine for thirty minutes. (durative) (2) Bill ate an apple/drank a glass of wine in five minutes. (perfective) In (1), the objects are indefinite and non-quantized, thus constituting activities. The objects in (2) are indefinite but quantized and makes a big difference in aspect. It denotes one whole apple, i.e., number (one here) plus atomic individual. It is an Incremental Theme (Dowty 1991) or a quantized Theme (Krifka 1987, 1997). When the predicate is telic in its semantic nature, it involves a homomorphism from its Theme argument denotations into a domain of events. If it is quantized, with numeral expressions or definite descriptions, with full individuations, then no proper subevents (of eating subparts of an apple) of an event of eating an apple can be described by the same sentence that includes an apple or other quantized expressions. Apples and wine are not quantized and are both massive. The bare nominal apples is also treated as a mass (cumulative) in contexts like three pounds of APPLES. The indefinite article, however, cannot be used for quantization in the following situations where the indefinite NP plus a light verb constitutes an activity. Consider: (3) a. I took a walk for thirty minutes. b. I took one walk in thirty minutes. In (3), take a walk as a whole is an activity (cumulative) and the indefinite article a is not used for quantization in this context, and, therefore, can be modified by a durative time adverbial (for---). A walk has no inherent bounds. Contrasted with a walk, one walk in (8), with an explicit numeral, can be quantized in event frequency; a quantized event nominal represents event frequency. Classifier languages including the Far Eastern languages have a frequency classifier for this kind of event nominals. Predicates showing the same behavior might include do a dance, make (a) noise, take a sip, etc. In no-article numeral classifier languages like Korean, a bare common noun has many faces: a mass noun face, a definite NP face, etc. In a cumulative situation like eat apples, as in (1), a bare noun occurs in Korean. In a quantized context like eat an apple, a numeral classifier construction occurs: (4) Mary -nun (samsip-pun tongan) sakwa -rul mok -oss -ta (5) Mary -nun (sam-pun man-e) sakwa han kae -rul mok -oss -ta Sentence (4) denotes an activity and cannot be modified by the in time adverbial, whereas (5) denotes an accomplishment and cannot be modified by the for time adverbial. If some completive auxiliary/adverb accompanies the VP, a bare nominal object in it is a (fuzzy) numeral or definite NP and therefore is quantized: (6) Mary -nun sakwa -rul (ta) mok-o pori/chiu -oss -ta Mary ate up the apple(s). (cf. *Mary ate up apples.) (7) Mary -nun sakwa se/myot kae -rul (ta) mok-o pori/chiu -oss -ta Mary ate up (the) three/a few apples. The combination of Numeral + Classifier can either follow or precede its associated common noun: (8) sakwa (???-tul) se kae (-ka/ -rul) (9) se kae -uy sakwa (-ka/ -rul) The post-N numeral construction (8) is predominant, i.e. more than 60%, and seems to be increasing in speech (Park and Kim 1996). Previously (Lee 1989) I proposed that the construction be derived from a modifying clause of [Subject N + Predicative Numeral Cl] preceding the head N. This predication treatment can be justified by the following facts: (a) Korean has the same kind of clause type, (b) there is (honorification) agreement between the head N and the classifier, (c) the classifier selects a possible head N and the violation of selection restriction creates graded sortal incorrectness: (10) yoca -ka se myong -i -ta [Subjrct-Predicate Relation] (11) sonsaengnim tu pun / ?*myong [Honorification Agreement] (12) a. ai se myong / *tae (machine class) [Selection Restriction] b. mal tu mari / ?*kae (small object class) c. kkangphae se myong/ ??mari (animal class) There is a degree of acceptability with mismatches between classifiers and head Ns; people applied to a machine classifier as in (12a) is a serious violation. (12b), with kae, is better: it is used by children to refer to anything, as a general classifier. The use of mari in (12c) has the metaphorical implicature involved: Gangsters are animals. The predication treatment of the Num Cl construction can handle such discrepancies between Cl and head N as violation of selection restriction, as shown in (12) above. The NP containing its internal post-N numeral classifier as in (12) is rather specific in information status. If the head noun is definite explicitly or contextually and the numeral classifier part is specific by being appositive to the head noun, the Num Cl cannot float out of the whole NP. (13) ku chaek tu kwon (-i/-ul) the/those two books (14) ku chaek-ul tu kwon sa-ass-ta two copies of the book (same type/subspecies) (15) ku sul-ul tu can / ku sul tu can-ul mashi-oss-ta two glasses of the wine. The head noun can be definitized (and therefore quantized) and its part can be quantized by numeral classifier expressions. The relation, then, can be either appositional (equal) or part-whole. In other words, the part can be the whole or a proper subpart of the whole. If the head noun is definite (or specific) likewise in any language, numeral quantization is subpart quantization, i.e. partitive. The head noun is plural-marked in non-classifier plural-marking languages like English. Atomic individuals are paid more attention (as in three apples). However, in classifier languages, the head noun is bare and normally singular-marked, being regarded more as mass/kind than as atomic individuals or their groups. We can have not only sakwa se kae but also sakwa se sangca and sakwa se kwan. The latter two have the sense of mass because of the measure phrases. But they don't necessarily denote pure apple meat contained. The only difference is that countables have their own inherent individuating bounds. We can exploit this feature by asking How many apples are in the three boxes? in Korean, English and other languages. Rice is mass if its grains on some occasion are not counted or measured by weight or bulk units. As long as they are not quantized, they behave as mass and form atelic or nontelic activity or process when associated with verbs in aspect and event structure. Therefore, Chierchia's (1996, 1997) suggestion that mass nouns come out of the lexicon already pluralized (furniture, water) might be Indo-European-based. If not quantized, things are mass, even if they are mereologically structured. Some classifiers can denote kinds and events. Some additional examples: (16) a. Individuals: kae [small-sized objects], mari [animal], myong [person], thol [grain] b. Measure: li [distance], kwan [weight], sangca [box], pyong [bottle], caru [sack] (17) Kinds: kaci [kind], congryu, cong [kind] (18) Events: pon [frequency], thonghwa, thong [phone call], hoe [frequency], kkon [achievement, love affair], thang [bad deed as a goal], thok [treat], charae [turn] The taxonomy of a common noun in Chinese that can be associated with different classifiers was illustrated by Ahrens and Huang (1996). The example telephone with similar semantic types can be found in Korean: (19) conhwa tu tae/kaci/thonhwa Event types can be predicted by the telic (purpose) feature of the original artifact telephone (Pustejovsky 1995). The event numeral classifier thonghwa in (19) simply reflects the frequency of the arbitrarily bounded events. So, we can further quantize the spatio-temporal restrictions by means of time and distance units, e.g. (conhwa) samsip-pun thonghwa 'thirty-minute call' or cang-kori conhwa 'long distance call', which constitute sub-kinds of the event classifier thonghwa. The expression han kkon olli-ta 'succeed in a pick-up/round-up' is often used to denote one successful pick-up or round-up. The kind-individual relation is not easy to capture and numeral classifier quantization is normally based upon common-sensically or folk taxonomically homogenous intra-kind/species, not upon heterogenous inter-kind/species membership. For instance, even if the classifier mari applies to all the kinds of animals ranging from a bacteria to a horse, we do not apply the numeral classifier [Numeral mari] to a mixture of cows and cats and bacteria. We apply it to cows, cats, or bacteria, separately. So, we can say that paekma yol mari '10 white horses' and mal yol mari '10 horses' are quite different. The latter can include horses belonging to different subspecies of the same intra-horse species, but not the former. This way, Kung-sun Lungs paradoxical pai ma fei ma may be justified in one reading. The classifier may be represented as: (22) mari = n x #(x) = n animalsubkind(x) (cf. Krifka 1997) I once gave a Korean example of kind-individual interaction. To repeat: (23) yoca -ka ta ssirum -ul ha-ne Even a woman is wrestling, I am surprised. The surprise marker can be applied to the perception of the individual instance of a woman wrestling and the exhaustion marker can be applied to the woman kind as a whole, say in contrast with man. We can apply some type shifting to handle the relations among e(individual extension), GQ<<e,t>,t>, and <e,t>(predicative), as suggested by Partee et al (1987) and Chierchia (1997), but without consideration of Topic-Comment information structure, it cannot be as fruitful as expected. Let us consider the following examples (Lee 1996) (24) tu namnyo -nun nul haengbokha -ta The two, man and woman, are always blessed. In the sense of any couple in (24), if the condition of having a couple of man and wife is met generically, the couple is always blessed. The relation between two and man-woman is appositional here. There is an arbitrary coupling operation involved for this Topic sense. Because of the conditional sense, some quantificational or modal expression usually accompany the utterance. For instance, (24) can have haengbokha-ke marion-i-ta is destined to be blessed as its predicate instead. Lis (1996) examples in Chinese that require modals are all this type of Topic constructions. Otherwise, if a numeral classifier expression gets the topic marker nun in Korean, whether sentence initially or mid-sentencially, the expression necessarily becomes a Contrastive Topic because of the focal force on number expressions. The question arises as regards to whether the same quantization constraint on complement NPs of an aspectual verb like begin as found in plural-marking languages (Pustejovsky and Bouillon 1994) applies to article-less languages with numeral classifier constructions and definite NPs with a zero-form definiteness. Consider: (26) Joe -nun pap -ul manhi mok -ki/ sul -ul manhi masi-ki sicakhae-ss -ta (27) Joe -nun sul/tambae/sosol -ul sicakhae-ss -ta (28) haksaeng-tul -un sul han/tu pyong -ul sicakhae-ss -ta In (26), the begin to V expression takes an indefinite (non-quantized) NP as the object of the verb, with a modifying vague amount expression. In this case, the interpretation of the whole VP is generic or habitual. The VP cannot represent a particular event. In (27), the object NPs can be either zero-form definite or indefinite generic and in their latter interpretation the Agent begins the respective activity for the first time. In (28), the numeral classifier NP constructions as objects get their quantization interpretation only of a particular event. We could see particularly how the quantization
function of numeral expressions is crucial in event, aspect and
argument structure. Aspectually, its delimiting or measuring out
function is crucially correlated with telicity. On the other hand,
informationally, information on number is rather novel and tends
to lean toward the verbal part rather than toward the Topic part.
컴퓨터에 의한 국어 "하"-형 용언의 분석과 생성 이 기 용 (고려대학교) 국어에는 어근이나 명사, 부사, 또는 형용사 어간이
용언화 접미사 "하"와 결합하여 동사나 형용사가 되는 매우
생산적인 과정이 있다. 예를 들면, "따뜻-하-다", "일-하-다"와
"가까-이-하-다", "좋-아-하-다" 등이 이러한 과정을
통해 생성된 용언들이다. 이 논문은 이같은 "하"-형 용언들을
컴퓨터 프로그램을 통해 자동으로 처리할 수 있는 규칙기반의 어형 생성
시스템 코몰(Korean Morphological system)을 구축하는 것이 그 목적이다.
◆ 1997학년도 제2학기 연구
발표회 계획 (한국언어정보학회): 제
35호 일시: 9/20 (토) 9:30, 10/20 (월) 18:00, 11/15 (토) 14:00, 12/13 (토) 9:30 장소: 대우재단 빌딩 (서울역 앞 대우 빌딩 뒤) 세미나실 (단, 11월 15일은 조선대에서) 발표 시간: 기획 논문/강의--60분, 일반 논문--40분
9월 20일 (토), 9:30 a.m. 기획: 윤영은 (이화여대) "Interpretations of Dependent Variables in Donkey Sentences" 일반: 이민행 (연세대) "자유초점 구문에서의 초점핵 결정원리에 대하여" 일반: 이예식 (경북대) "Are Negative Polarity
Items Really Licensed?" 10월 20일 (월), 6:00 p.m. 기획: Steven Lapointe (UC-Davis) "Gerund Phrases in VSO Languages: Welsh and Maori" 기획: 윤혜석 (서울대) "Mixed Categories and
the Lexical Integrity Principle" 11월 15일 (토), 2:00 p.m. (장소: 조선대학교) 기획: 강정구 (KAIST) "관계적 관점으로서의 시제와 상: Klein의 이론과 응용" 기획: 백미현 (충남대) "The Korean I-suffix on 'Passive' Constructions" 일반: 류병래 (서울대) "Middles in the Constraint-based Lexicon" 일반: 염재일, 이익환 (연세대) "Common Grounds as Multiple Information States" 일반: 장석진 (서울대) "번역의 대응과 평가
기준" 12월 13일 (토), 9:30 a.m. 기획: 유은정 (서울대) "Wh-interrogatives and a Theory of Quantification" 일반: 홍민표 (명지대) "A Pragmatic Analysis of Quantificational Force in Wh-phrases" 일반: 채희락 (한국외대) "서술명사의 정체성과 통사적 정의 기준" 일반: 이정민 (서울대) "Numeral Classifiers and Quantization" 일반: 이기용 (고려대) "컴퓨터에 의한 국어
"하"-형 용언의 분석과 생성" # 대우재단 빌딩의 주차장이 협소하고 주차권 발급이
한정되어 있사오니 (5매), 가급적 대중 교통 수단을 이용하여 주십시오.
|