문어 텍스트와 구어 텍스트의 담화 표지와 텍스트 구조 비교 연구

이숙의(Suk-eui Lee),김진수(Jin-soo Kim) 어문연구학회 2008 어문연구 Vol.56 No.-

This study is focused on the comparative analyzing between the literary expression and the oral expression, but those two texts belongs to same prototype. One of the research material of this study is Oakdanchoonjeon which is a Korean classical novel, the other is a Lee Hyunryong and Kim Jinhoi which is an oral narration of Oakdanchoonjeon. Firstly discourse markers of two texts were compared. Most of cohesive devices in Oakdanchoonjeon are grammatical, logical and semantical binders. However, in Lee Hyunryong and Kim Jinhoi, pragmatical and internal binders are remarkable, like "ah, in-ja, mak". Secondly, each text structure was analyzed and compared. On the case of literal text, it was focused on the success story of Lee Hyunryong. On the contrary, in the oral text, it was focused on the love story between Oakdanchoon and Lee Hyunryong. It shows that a text structure of oral text can be changed by speaker"s focusing, narrative place and time. Also discourse markers are related to the episodes, it functioned as a cohesive device between episodes. Especially, using discourse markers was remarkable in oral text.

국어 용언의 품사 분류에 관한 인지적 접근

이숙의(Suk-eui, Lee) 어문연구학회 2012 어문연구 Vol.72 No.-

본고에서는 ‘늙다’와 ‘모자라다’와 같은 품사 경계가 모호한 어휘에 대한 인지 언어학적 의미 해석을 시도하고, 연속적 접근(continuum approach)관점에서의 품사 분류 방법을 논의한다. 따라서 동작 동사와 상태동사의 범주 경계 모호성에 관한 인지적 접근 방법을 통한 품사 분류 관점의 도입을 제안하였다. 구체적 기술을 위한 논의로, 4장에서는 ‘가다, 늙다, 모자라다, 젊다’의 어휘의미와 상표지의 결합 관계에 대해 중점적으로 기술하였다. 이를 통해 연속적 접근방법을 이용하여 범주의 경계에 있는 모호한 어휘의 품사 적용에 대한 타당한 설명을 하고자 시도하였다. 동사와 형용사의 범주에 속하는 구성요소들 사이에는 정도나 등급의 차이가 있는 여러 유형이 있다는 인지언어학적 전제 아래, 과정의미가 없고 시간성과 무관한 어휘라도 탄도체와 지표의 일치에 대한 과정적 인식과 같은 해석이 가능하다면 동사범주에 속한다고 보았다. 본고의 의의는 형태, 의미, 기능의 세 가지 기준에 의한 이분법적 분류만으로는 납득하기 어려운 어휘에 대해 인지적 차원의 연속적 접근 관점으로 품사의 적용에 대한 새로운 해석 방법을 제안한다는 것에 있다. 여기에서는 두 범주에 속하는 네 가지 유형의 어휘에 관한 논의에만 한정하였으나, 상황에 따라 동사와 형용사로 통용이 가능한 다른 유형의 어휘에 관한 의미 분석이나 다른 품사 범주의 경계에 놓인 어휘에 관한 연구의 연장된 논의가 필요하다. The purpose of this paper is examining a predicate class of Korean from cognitive linguistical viewpoint. In prescriptive grammar of Korean, predicate category is divided into two parts, verb and adjective class. Some predicate has properties both verb and adjective at the same time, it is difficult to decide their parts of speech, such as ‘늙-, 모자라-’. Focus of here is on the explaining why a word belongs to the verb or adjective class by analyzing cognitive meaning. For example, ‘모자라-’ does not relate to [process] and [time], but it isn’t an adjective. To categorization, the continuum approach is more useful than the dichotomy approach. The subjects of the analysis are four words. ‘가-’, ‘젊-’ are typical examples of each class, ‘늙-’, ‘모자라-’ are the marginal examples. ‘가-’ is belong to the verb class; this is obvious fact, because the lexical meaning has [process] and related to [time], so it is natural the stem of ‘가-’ can combine with ‘-ㄴ/는-’, ‘-고 있다’ and ‘-어 가다’. However, ‘늙-’ is little different case. [Process] meaning of ‘늙-’ is abstract, but it is related to [time], because the movement of the traject is abstract and the domain of landmark is unclear. From movement of traject, although movement of traject is invisible, ‘늙-’ can combine with ‘-ㄴ/는-’, ‘-고 있다’ and ‘-어 가다’. ‘모자라-’ doesn’t relate to [process] and [time], means [state]. It has no movement of traject in schema. But landmark has standard point. It is a kind of destination (or directing points) where traject proceed to reach. A gap between traject and landmark area relate to change of “degree of lack”, so it become more similar to verb. ‘젊-’ means [state] and [non-time], also there is no standard point inside landmark as ‘모자라-’. A a gap of area between traject and landmark does not relate to change of “degree of young”. In this respect ‘젊-’ is obviously different with ‘모자라-’, and then it can not combine with present or progressive marker.

온톨로지 지식과 개체명 세분화를 통한 맥락의미 자동인식의 향상

이숙의(Lee, Suk-eui) 한국언어문학회 2016 한국언어문학 Vol.99 No.-

The named entities(NE) of the special domain to be covered in this paper is the concept category name of the vocabulary. Ontology integrates the conceptualization of the derivation and expansion of lexical semantics. Therefore, the ontology data are suitable for understanding the meaning of vocabulary according to context. The purpose of this paper is to utilize ontology knowledge in the construction process of NE learning data of special domain that can be used in automation system for query response. This paper proposed a method to improve the accuracy of NE recognition in automatic natural language processing. [Child Care] domain document and consultation dialogue data was constructed, and tried to find out how to classify the NE necessary for automatic classification of documents and Q & A system. In Section 2, the necessity of subdivision of NE category and similarity with ontology were introduced to understand object name recognition through context information. Chapter 3 introduces the collection of [Childcare] documents and describes the results of analysis of specific vocabulary use. In chapter 4, how to refine the NE in the [Childcare] documents was shown and illustrated the necessity of classification. As a result, the accuracy rate of recognition of the object name in all 6000 sentences was over 95%.

부사성의 강화와 부사화

이숙의(Lee Suk-eui) 어문연구학회 2016 어문연구 Vol.88 No.-

본고는 ‘부사화’라는 용어의 근본 개념에 충실하며 그동안 부사화로 다뤄진 어휘들이 진정다른품사에서부사로품사가전성되어사용된다고밖에볼수없는것인지에대하여 숙고해 보고자 하였다. 즉 그간 부사화 되었다고 본 어휘들이 타 품사범주에서 부사의 범주로 유입된 것인지, 아니면 원래 부사였을 가능성은 없었던 것인지에 대해 살펴보고자하였다. 여기에서는 부사화 되었다고 보는 어휘들 중 일부가 다른 품사에서 부사로 바뀐것이 아니라 원래부터 부사였을 가능성이 있다는 관점에서 부사화와 부사성 강화를 대조적으로 설명하고자 하였다. 이와 관련하여 먼저 품사분류에 관한 연속적 접근 방식을 통해 명사와 부사 범주 경계의중첩가능성을제시한다. 또두범주의문법적속성을동시에지닌어휘가상황에따라 부각되는문법적속성에의해두범주중하나에속할수있음을기술하고자하였다. 이는 기존의 품사통용에 관한 설명과 유사하다. 그러나 속성의 부각과 활성화를 통해 설명하고자 하였다는 점이 기존의 품사통용에 관한 논의와는 다르다. 그간 단어 형성 관점에서 다뤄졌던 어휘의 품사전성에 관한 논의에 대해 인지언어학적 관점을 도입하여 어근의 부사성 강화를 부사화와 구분함으로써 부사형성에 관한 새로운시각으로 접근하고자하였다. 어근이가진여러문법적 특질의활성화에따른품사성의 부각으로 동일 형태의 다품사성을 설명하는 것은 그간 품사전성이나 통용이라는 개념으로 설명하기 어려웠던 문제를 해결하는 데 새로운 시각을 제시한다는 점에서 의의가 있다. This paper is about ‘the adverbial reinforcement' and ‘the adverbalization'. It tried to investigate some words that were covered as a result of adverbalization for a long time. Especially, topic of this paper began with distinction between adverbalization and adverbial reinforcement. This paper focused on the view of adverbial reinforcement. At the view point of cognitive linguistics and fuzzy grammar theory, word class(category of the part of speech) depends on the other word class. Most categories share the part of intersection. From the concept of ‘continuum approach' of cognitive semantics, boundary of the word class is fuzzy. Moreover each constituents(words) of same class don't have same qualification. Some words are placed on the center of class but some words are placed on the boundary of class. Some of predicative nouns which is referred to Sino-Korean words shows to strong conversion tendency, a word could belong to multi classes. These kinds of predicative nouns are studied as a perspective of adverbalization. However, root of predicative nouns have not only grammatical features of noun but also have features of determiner and adverb. In this paper, the phenomenon that the root of predicative noun which is used as a adverb was not taken as a case of adverbalization, but it was described as the adverbial reinforcement. Until now, the word class(category of the part of speech) was from the point of formation of word. This paper applies the word class(category of the part of speech) at the view point of cognitive linguistics and fuzzy grammar theory. With this new perspective, this paper explains the phenomenon that the root of predicative noun was not taken as a case of adverbalization, but it was described as the adverbial reinforcements. For many years, it had difficulty explaining many parts of speech with the word class(category of the part of speech) and commonly used concepts. This paper is meaningful because it gives new perspective of explaining those concepts in the view point of cognitive linguistics and fuzzy grammar theory.

높임 선어말 어미 '-시-'에 대하여

이숙의(Suk Eui Lee) 한국어학회 2015 한국어학 Vol.66 No.-

'-Si-' is often known as a Korean pre-final ending which is a honorific marker for a subject(theme) in a sentence, it is rather grammatical. In recent years, many researchers say that `-si-` is a kind of pragmatical maker to respect for ``hearers``. This paper acknowledged both opinions, the former is dealt with core meaning and the latter is accepted as peripheral meaning. Appearance of `-si-` in a sentence is very usual phenomenon in general utterance. This phenomenon of using `-si-` started with intention of speakers, and it is a disagreement between subject and honorific marker. Focus of this paper is on the sentence production process including `-si-`, and why this has happened. Previous studies concentrated more on the meaning and function of `-si-`, and tried redefinition of `-si-` grammatically or pragmatically. However, it is impossible to denominate one so-called marker in the respect of numerous grammatical disagreement phenomenon of `-si-`. `-Si-`is both grammatical and pragmatical marker, which has basic meaning of [honorific]. Simply this paper focused on a fact that it is proceeded from speaker``s strong strategy, and tried to attest to the fact by using Garrett(1975, 1976)`s evaluation on the speech error. If a speaker plans to respect a hearer, he(or she) intensively induces a disagreement phenomenon between subject and honorific pre-ending of predicate, even though he(or she) knows that is wrong. Regardless of the fact that `-si-`is a honorific maker for ``the subject` or `the hearer` in a sentence, using `-si-` is completely dependent to conceptualization layer which is message level in the model of speech production.

신문 기사 분석을 통한 ‘코로나-19’ 어휘 연구

이숙의(Lee, Suk-eui) 한국어학회 2021 한국어학 Vol.93 No.-

본고에서는 최근 사회적으로 큰 이슈가 되고 있는 코로나-19에 관한 사회적 담론의 생산, 발전, 변화의 과정을 살펴보고자 하였다. 본고에서는 일반 명사와 동사를 중심으로 고빈도 어휘의 사용 양상과 의미 부류별 분석을 진행하였다. 그 결과 [방역], [의료], [교육]과 [경제] 위기에 관한 사회적 관심이 매우 높다는 점, 질병의 [발생]과 [증감], [통계] 정보가 신문을 통해 많이 유통되었음을 알 수 있었다. 또한 개체명 인식 결과 코로나-19의 의미적 특수성을 반영한 직위/직책명(CV_POSITION)과 나라명((LCP_COUNTRY), 도시명(LCP_CITY), 도, 주 지역명(LCP_PROVINCE), 발병 원인과 관계가 깊은 종교단체명(OGG_RELIGI ON)과 신체 조직 및 기관의 명칭(TM_CELL_TISSUE) 등의 출현 빈도가 높았다. 본 연구는 코로나-19의 발생 시점부터 현재(2021년 8월)까지 신문에 사용된 어휘를 형태소 분석 결과별, 개체명별로 폭넓게 분석하고자 하였다는 점에 의의가 있다. In this paper, the process of production, development, and change of the COVID-19 discourse, which has become a big social issue, was examined through lexical analysis. This study is meaningful in that it attempted to deal with the use of vocabulary from the first occurrence of COVID-19 to the present(Aug, 2021). Patterns of use of high-frequency vocabulary and semantic classes were analyzed focusing on common nouns and verbs. As a result, social interest in the [prevention] and [medical], [education] and [economic] crises was very high. In addition, it was found that a lot of [occurrence], [increase/decrease], and [statistics] information of diseases were distributed in newspaper articles. As a result of NER(Named Entitiy Recognition), the frequency of occurrences of words belonging to title/position name (CV_POSITION), country name (LCP_COUNTRY), city name (LCP_CITY), state name (LCP_PROVINCE), religious group name (OGG_RELIGION), and body organization name (TM_CELL_TISSUE) was high. All of those vocabularies reflected the semantic specificity of COVID-19.

[보험계약상담대화]프레임의인지의미론적연구

이숙의(Lee Suk eui) 어문연구학회 2017 어문연구 Vol.92 No.-

본 연구는 틀 의미론과 인지의미론 관점에서 보험계약 상담대화 프레임의 구성요소와 하위사건구조를 분석하고, 하위사건의 흐름에 따른 구성요소의 활성화 양상을 분석하여 [보험계약상담대화]프레임의개념기술을시도하고자하였다. 전체[보험계약상담대화]프레임의 하위사건프레임에 핵심구성요소가 어떻게 관여하는가를 살펴보고, 하위사건프레임과 핵심구성요소프레임의 통합을 통해 전체 [보험계약상담대화]프레임의 개념과 상황정보에관해종합적인고찰을시도하였다. Fillmore의틀이의미적틀에한정되어발화사건과독립적인 지식 목록을 기술한 것이라면 본고의 대화프레임에 관한 설명은 담화문맥에 틀을 부여하는발화사건틀(speech event frame)에관한분석이라는점에서다르다. 발화사건틀은 상황에 관한 지식을 도식화한다. 따라서 어휘목록과 문법지식에 관한 배경지식에 관한 이해를 넘어 상황정보에 관한 지식을 도식화한다. 본고는 어휘지식표상 단계의 차원을 넘어 사건프레임에 관한 상황정보표상에 관한 기술을시도하였으며, 발화사건틀에관한인지언어학적접근을시도하였다는점에의의를둔다 This paper tried to define the attributes of the insurance contract consultation dialogue frame in conceptual relation with other frames. The typical structure of the frame was established through dialog analysis. It was analyzed the sub-event structure and development direction of the whole frame and sub-concepts of core components, was examined whether the sub-event frame is appropriate for the purpose of the conversation. The vocabulary usage and the frequency were used to identify the activation patterns of core components according to the purpose of the sub-event frame. Therefore, the degree of activation of key components in the subcases of the insurance contract consultation dialogue frame was grasped. Fillmore s framework(1982, 1985) is limited to a semantic framework, which describes a fictional event and an independent knowledge list. The description of the dialogue frame in this paper differs from Fillmore s semantic framework in that it analyzes the speech event frame that gives a frame to the discourse context. The speech event frame schemes the knowledge of the situation. Therefore, it is more than understanding of vocabulary list and grammar knowledge. It can be extended to studies on human experience and conceptual cognitive systems. The core components of the insurance contract consultation dialogue are related to [economy], [contract], and [health]. The middle stage of [Insurance Contract Consultation Dialogue Frame] consists of five stages of E1-E5. E1 and E2 are the preliminary stages, and E4 and E5 are the overall stages of the event. E3 is the most typical step of the consultation frame. In E3, the vocabulary is the most diverse. It can therefore be further subdivided into other stages. [Insurance contract consultation dialogue] differs depending on language and culture. The dialogue structure differs according to personal experience. This paper uses the example of insurance contract consultation conversation as a unit. Through the dialogue structure and lexical analysis, this paper tried to describe the situation information representation of event frames. And that the cognitive linguistic approach to dialogue analysis was attempted.

딥러닝을 활용한 Simon 블록 암호 분석에 대한 연구

성효은(Hyoeun Seong),김예원(Yewon Kim),강주성(Ju-Sung Kang),염용진(Yongjin Yeom) 한국통신학회 2020 한국통신학회 학술대회논문집 Vol.2020 No.8

유사 문장 말뭉치 분석을 통한 유사도 인식에 관한 연구

이숙의(Lee Suk-eui) 어문연구학회 2021 어문연구 Vol.108 No.-

본고에서는 유사 문장 말뭉치를 대상으로 자동시스템과 사람의 유사도 인식 결과를 대조·분석하였다. 자동시스템과 사람의 유사도 인식 불일치 결과를 주로 다뤘는데, 그 이유는 이러한 연구를 통해 컴퓨터와 인간이 문장을 이해 하는 방식이 서로 어떻게 다른가를 보기 위함이었다. 이러한 연구는 추후 구축될 유사 문장 말뭉치와 유사 문장 자동처리시스템 개발에 활용할 수 있다. 본 연구는 한국전자통신연구원의 유사 문장 학습 데이터 구축 결과를 활용하였고, 구축 과정에 적용한 유사도 판단을 위한 기준에 관한 검토를 진행하였다. 6개 구간의 점수대로 유사도 문장 쌍의 점수 값을 부여하였고, 3점 이상의 문장은 유사 문장에 해당된다. 주로 유사 문장의 경계에 놓인 문장 쌍들이 컴퓨터와 인간의 유사도 판단의 차이가 있는 지점임을 알 수 있었다. 자동시스템에서는 유사 문장으로 판단하지만 사람은 유사 문장이 아니라고 판단 한 예들을 중심으로 살펴 보았다. 이 결과를 통해 기초 작업에서 모호성이 짙었던 유사도 판단 기준을 구체화 하였으며, 이를 추후 연구에 반영하고자 한다. This paper analyzed the corpus of similar sentences and compared the results of the computer system and manual work. In order to analyze how computers and humans understand the meaning of sentences differently, the results of inconsistency between humans and automatic systems for recognizing similarity were mainly analyzed. This paper utilized the results of constructing similar sentences learning data of the Electronics and Telecommunications Research Institute(ETRI). Here, the similarity criteria applied in the process of constructing language data for the ETRI project were reviewed. A similarity score was given by dividing into 6 sections, and sentences with a score of 3 or more are similar sentences. In Chapter 2, the definition of linguistic similar sentences was presented, and similar sentences types were classified. In addition, sentences surveyed in the actual corpus were also presented. Chapter 3 introduced the process of constructing a corpus of similar sentences. the results of corpus construction were used in this study. In Chapter 4, an example of inconsistency between the results of the automatic system and manual work was analyzed. In addition, I tried to reveal the difference between the automatic system that operates based on rules and the way humans understand sentences. It is hoped that the results of these studies will be utilized in the future development of the automatic processing system for similar sentences and the direction of corpus construction.

온톨로지 지식과 개체명 세분화를 통한 맥락의미 자동인식의 향상

이숙의(Lee, Suk-eui) 한국언어문학회 2019 한국언어문학 Vol.109 No.-

The named entities(NE) of the special domain to be covered in this paper is the concept category name of the vocabulary. Ontology integrates the conceptualization of the derivation and expansion of lexical semantics. Therefore, the ontology data are suitable for understanding the meaning of vocabulary according to context. The purpose of this paper is to utilize ontology knowledge in the construction process of NE learning data of special domain that can be used in automation system for query response. This paper proposed a method to improve the accuracy of NE recognition in automatic natural language processing. [Child Care] domain document and consultation dialogue data was constructed, and tried to find out how to classify the NE necessary for automatic classification of documents and Q & A system. In Section 2, the necessity of subdivision of NE category and similarity with ontology were introduced to understand object name recognition through context information. Chapter 3 introduces the collection of [Childcare] documents and describes the results of analysis of specific vocabulary use. In chapter 4, how to refine the NE in the [Childcare] documents was shown and illustrated the necessity of classification. As a result, the accuracy rate of recognition of the object name in all 6000 sentences was over 95%.