한국어 의미역 말뭉치 구축을 위한 의미역 표지 연구

신서인 한국어의미학회 2019 한국어 의미학 Vol.66 No.-

The purpose of this paper is to discuss what should be considered in the process of establishment of the semantic role label set and semantic role annotation for a semantic role annotated Korean corpus. This paper first compares the semantic role label lists of dictionaries or corpora that are built on a large scale, and examines the status of the semantic role annotation, and examines the issues related to the semantic role annotation of arguments. Accordingly, semantic role labels for adjunct are proposed. For the practical purposes, such as the establishment of an automatic response system, it would be advisable to use the same semantic role set to identify the agreement between the two sentences, but the current paper proposes the semantic role annotation suitable for expressing the relationship between the predicate and the arguments in a sentence. Decisions on how to solve the semantic annotation problems of Korean sentences should be made before a balanced list of semantic role label sets and clear work instructions can be drawn up.

存現句賓語的題元角色分析

황후남 ( Hwang Hoo Nam ) 한국외국어대학교 중국연구소 2017 中國硏究 Vol.70 No.-

This paper aims at Modern Chinese to analyze the thematic role of the object in existential sentence, and the results are as follows: 1) What is the Semantic Role of the Object in Existential Sentence? In the view of most of the traditional grammars, the semantic role is Patient when the object of existential sentence in Modern Chinese is inanimate, otherwise, the semantic role is Agent; however, generative grammar is on the point that the semantic role is Patient no matter the object is inanimate or animate. This paper agrees with the latter and believes that in the existence sentence of Modern Chinese, the semantic role of the object is Patient. 2) How Many Arguments Are There in Existential Sentence? No matter the sentence is stative or dynamic, existence sentence or fugitive sentence, and no matter the object is inanimate or animate, there are three arguments in one sentence, that is, internal argument, middle argument and external argument. 3) How Many Hierarchies Are There to Form the Existential Sentence? The first one is the outside event hierarchy whose external argument is pro and gets the semantic role of Causer from light verb CAUSE; the second is the inside event hierarchy in which the place is the middle argument and gets the semantic role of Affectee from light verb BECOME; the third is basic vocabulary hierarchy whose object is the internal argument and gets the semantic role of Patient from unaccusative verb. 4) How Many Eventualities Are There to Form the Existential Sentence? The whole sentence indicates the empty subject and causes the happening of certain event which makes the place change, and this change leads to the appearance of the person or thing that the object stands for. 5) What Does the Pro Refer to in Existential Sentence? No matter in existential sentence with inanimate object or existential sentence with animate sentence, there is pro in subject place and the pro is the Causer. In existential sentence with inanimate object, the “empty subject” assumed in this paper is the external argument, and the Causer caused “predicate + object”. The “empty subject” here means the Causer that only can be found outside the sentence instead in the sentence. In existential sentence with animate object, the “empty subject” is also the external argument, even though the “empty subject” is also the Causer caused “predicate + object”, it can be found what it refers to in the sentence, that is, the inverted sentence of the original sentence “predicate + object”. The inverted sentence is the subject-predicate structure, and the light verb of the predicate is DO.

한국어 의미역 기술 모형 개선을 위한 연구 - Korean PropBank 프레임파일의 구조 개선을 중심으로 -

신서인 한국어의미학회 2020 한국어 의미학 Vol.70 No.-

The purpose of this paper is to analyze the structure of the Korean PropBank framefiles and suggest several ways to improve it. In the paper, we discuss the features and limitations of the Korean PropBank framefiles and propose an alternative model for semantic role description. In terms of macro structure, the Korean PropBank framefiles have a peculiar structure. The lemma representing several words becomes the entry, and the related vocabulary is described in one file. Therefore, the lemma setting criteria and related vocabulary processing criteria should be clearly established, and homonyms should be processed in a consistent way. In terms of micro structure, there is no place in the Korean PropBank framefiles to describe postposition(josa) information, which is essential for Korean semantic role labeling. Therefore, this lack has to be supplemented. In describing the sematic role information, it is more useful to describe it as a label-type sematic role rather than as a number-type sematic role or a narrative-type sematic role. In the Korean PropBank framefiles, arguments and adjuncts are analyzed in different ways in the analysis of example sentence, but they need to be analyzed parallelly. In order to take care of the suggestions for improvement, we propose an alternative model for Korean semantic role description. This new model will be able to describe information that is useful for a more effective analysis of Korean semantic role.

부가어의 의미역에 대하여

이홍식 ( Yi Hongshik ) 숙명여자대학교 한국어문화연구소 2020 한국어와 문화 Vol.27 No.-

본고에서는 부가어의 의미역을 설정하는 방법을 논의하고 구체적인 부가어의 의미역 목록을 제시하였다. 의미역은 전통적으로 논항이 술어와 맺는 의미론적인 관계를 의미하였다. 그러나 최근에는 부가어와 술어가 맺는 의미론적인 관계를 표상하기도 하고 문장의 의미에 문장을 구성하는 성분들이 기여하는 의미를 표상하기도 한다. 본고에서는 기본적으로는 술어와 문장 성분이 맺는 의미론적 관계를 의미역으로 이해하면서도 나아가 문장의 전체 의미에 기여하는 각 성분의 의미론적인 역할을 의미역에 포함시켰다. 논항과 부가어는 필수성의 측면에서 차이를 보이기 때문에 근본적으로 논항의 의미역과 부가어의 의미역이 구별되지는 않는다. 다만 결과적으로 어떤 의미역은 논항에만 나타나고 어떤 의미역은 부가어에만 나타나게 된다. 이러한 구별의 근본적인 원인에 대해서는 추가 논의가 필요할 것이다. 본 논문에서 확인한 부가어의 의미역은 모두 19개였다. 거리, 경로, 기간, 기준, 도구, 동반자, 목적, 방법, 방향, 범위, 비교기준, 시간, 양상, 원인, 자격, 장소, 재료, 정도, 조건. This thesis explored how to classify semantic roles of Adjuncts. Tradi-tionally the semantic roles were classified using a method that relates the nature of the two arguments with their predicate. There are some papers in which semantics roles are alloted to adjuncts. If we admit that predicates have empty slots and arguments which satisfy the semantic relations with the predicates fill the slots, we cannot allot semantic roles to adjuncts. But if we consider semantics roles as semantic contribution of sentence con-stituents to the meaning of sentences, we can admit that adjuncts have semantic roles. We can classify the semantic roles of adjuncts according to the relationship with their predicates or their semantic contribution to the meaning of sentences. We can identify some semantic roles of adjuncts by using this method; temporal, location, direction, path, period, criterion, qualification, extent, instrument, manner, purpose, condition. companion, cause, appearance. degree, material, comparison.

双宾构式语义特征对动词与论元角色的语义制约

林艳 중국인문학회 2011 中國人文科學 Vol.0 No.47

The article analysizes the semantic structure and the semantic feature of the ditransitive construction’s argument roles, applying construction grammar theory and metaphor theory。Taking up-down studying method, the article first summarizes the semantic structure of the ditransitive construction’s argument roles and it’s event structure on the basis of analysizing the ditransitive construction’s cognitive experience。 Second, the article analysizes every argument role’s semantic feature,probes argument roles’ semantic relations。 applying metaphor theory and metonymy theory, on the purpose of summarizing the argument roles’ semantic relations of the ditransitive construction and the matching between these argument roles, thus adding new evidence of applying construction grammar to study Chinese ditransitive construction。

어순 수용성 데이터 기반의 중국어 문두 다중논항 의미역 배열 기제 및 위계 연구

박정구 ( Park Jungku ),강병규 ( Kang Byeongkwu ),박민준 ( Park Minjun ),유수경 ( Yu Sukyong ) 한국중국언어학회 2023 중국언어연구 Vol.- No.109

This paper extracted a large number of sentences with multiple arguments at the sentence beginning position from various genres of modern Chinese corpus data through coding programs as well as artificial methods. After changing the order of arguments of sentence-initial position in the extracted sentences, word order acceptability data was constructed through verification by native speakers. For the large amount of data constructed in this way, the quantity and ratio of combinations of semantic-roles that the multiple arguments at sentence-initial position play were statistically analyzed. Based on the analysis so far, it has been possible to induce the constraints on the multi-argument arrangement and establish the hierarchy of the semantic role for topicalization. Among obligatory arguments that appear at the beginning of a sentence, the arguments used as the object precede the arguments used as the subject. Among optional arguments, Time, Locative, and Cause arguments precede other arguments. Source and Goal arguments have something in common: they tend to precede other arguments when they express time. However, when they represent locative meaning, they shows different characteristics. The locative Source follows the obligatory arguments, and the locative Goal precedes the obligatory arguments. Instrument, Manner and Material are closest to the predicate, and Beneficiary and Recipient precede obligatory arguments. The word order of multiple arguments at sentence-initial position is influenced by iconicity, semantic relationship with the verb, and whether or not they are obligatory arguments.

말뭉치 기반 중국어 문두 다중논항의 의미역 배열 경향성 고찰

박정구 ( Park Jungku ),강병규 ( Kang Byeongkwu ),박민준 ( Park Minjun ),유수경 ( Yu Sukyong ) 한국중국언어학회 2023 중국언어연구 Vol.- No.106

This paper examines the ordering patterns of multiple arguments preceding predicates in Mandarin Chinese. The focus of the study is to investigate how the order of multiple arguments at the beginning of sentences affects their acceptability, and to identify any constraints that may exist. To achieve this goal, the study extracted about 5,000 sentences from modern Chinese corpus data of various genres in which multiple arguments appear at the beginning of a sentence as topics, and conducted a statistical analysis of the word order arrangement of arguments according to different semantic roles. Through the analysis of this paper, the word order relationship between arguments with different semantic roles could be identified as shown below. First, Time/Locative precedes all semantic roles, and Time tends to precede Locative. This is a consequence of the fact that the Time/Locative functions as a theme providing a background for the information of the rest of the sentence. Second, it was confirmed that there was a certain tendency in the word order between obligatory arguments. In general, the semantic roles used as the subject before the verb in the SVO sentence are preceded in the topic sentence by the semantic roles used as the object after the verb. In the opposite case, it is accompanied by contrastive meaning and has strong restriction on acceptability. Third, the optional arguments as topic can be divided into Instument, Manner, Material, Causes, Reason, and Source. Instrument, Manner, Material and Spacial Source tend to follow obligatory arguments, and Causes, Reason, and Temporal Source tend to precede obligatory arguments. This can be explained by the restriction based on cognitive iconicity.

의미역 태깅의 제문제 고찰- 구축된 격틀사전의 문제점과 대상말뭉치의 문제점을 대상으로

김윤정 ( Yun Jeong Kim ),옥철영 한국언어정보학회 2016 언어와 정보 Vol.20 No.3

This paper deals with two problems occurring in the process of labelling semantic role to Sejong Corpus. The first problem is some discrepancy among a case-frame dictionary used for labelling semantic role, a constructed case-frame dictionary and Sejong corpus. The typical examples of discrepancy between a case-frame dictionary and Sejong corpus are ``appearance of compound words`` and semantic role tagging to the sentence with syntactic auxiliary verb constructions. To solve this problems, an extended case-frame dictionary should be constructed with supplement and modification of annotated semantic roles. The other problem is a verb problem not solvable with the semantic tagging program, ‘UTagger-SR’. Since ‘UTagger-SR’ generates and presents verbs as their original forms of presented predicates, compound words with space or syntactic auxiliary verbs are not taken into consideration. The worker tagging semantic roles should tag semantic roles, treating a compound word with spacing as one word, and a sentence with an auxiliary verb construction as one verb consisting of a non-auxiliary verb and an auxiliary verb. Then, in the process of generating results, post-processing is needed. The post-processing includes treating a verb belonging to compound words as a compound word regardless of spacing between letters and combining the auxiliary verb and the non-auxiliary verb together in case of an auxiliary verb construction. (Ulsan University)

부가어의 의미역 실현 양상 -TMP, LOC, EXT를 중심으로-

최형강(Choe, Hyeong-Gang) 한국어문학회 2020 語文學 Vol.0 No.150

본고는 ‘부가어의 의미역’ 중에서 시간, 장소, 범위를 표시하는 TMP, LOC, EXT에 초점을 맞춘 논의이다. 통사론적인 정보를 통해서는 ‘-를’만을 요구하는 서술어로 확인되더라도 의미구조에 따라 LOC, TMP, EXT를 선별적으로 요구할 수 있기 때문에 LOC, TMP, EXT는 부가어의 의미역에서 중요하게 다룰 필요가 있을 것이다. 또, LOC와 TMP, EXT의 관련성에 주목할 필요도 있는 것이다. 의미 정보에 주목하면 ‘-에서 -까지’ 전체에 EXT를 부여하는 것이 EXT의 사용을 좀 더 원활하게 하는 방법이 될 수도 있을 것이다. 그렇다면 TMP에 대해서도 동일한 방식으로 처리하는 것이 TMP의 분포를 한정하는 데에도 도움이 될 것이다. LOC의 경우는 TMP와의 관련성을 통해 부가어의 의미역으로서의 지위를 확고히 할 수 있을 것이다. LOC의 경우에는 논항의 의미역으로 상정될 가능성도 높지만, 부가어의 의미역으로 LOC가 부여되는 것도 배제할 수는 없을 것이다. This paper focuses on the semantic roles of adjuncts TMP, LOC, and EXT. TMP, LOC, and EXT need to be considered within the semantic roles of adjuncts because they can be selected according to the semantic structure even if the predicate requires only "-를" through syntactic information. It is also worth paying attention to the relevance of LOC, TMP and EXT. By considering semantic information, granting EXT to the entire "-에서 -까지" may be a way to make the use of EXT extensively. If so, treating TMP in the same way would also help to limit the distribution of TMP as a semantic role. In the case of LOC, its relevance to TMP ensures its status as a semantic role of adjunct. LOC is highly frequently used as a semantic role of argument, but it cannot be ruled out that LOC is granted a semantic role of adjunct.

국어의 목적어를 어떻게 정의해야 하는가?

한정한(Jeong Han Han) 한국어학회 2015 한국어학 Vol.68 No.-

The purpose of this paper is to define Korean object in terms of semantic features of transitivity that a predicate has, expecially based on generalized argument roles such as {O}, or {T} in linguistic typology. It is well acknowledged among Korean scholars that Korean object is difficult to define directly by formal approaches. This is because the object can realize without object case marker, and sometimes occur with particles like NUN(topic), MAN(``only``), TO(``also``) etc., that have their own semantic meanings. To define Korean object this way, we in this paper provide with typologically prevailed definition which was originated from Comrie(1978), and Dixon(1979), where object is defined by numerical valence and generalized argument roles. Following to them, Korean case follows Nominative-Accusative case frame where {S} of intransitive verbs and {A} of transitive verbs are marked the same. To make it further, we, based on corpus data, define Korean object as {O, T}. {O} means a most-patient-like argument of two place argument transitive verbs, and {T} means a most-patient-like argument of three place argument transitive verbs respectively. However, the definition like above could also have subsequent problems left out. For instance, because we did not define Korean object in terms of case marker LUL, it is natural consequence to answer what kind of functions or meanings that object marker LUL have, and how many types of transitive predicate are there in Korean. To answer it, we have examined the notion of transitivity of Hopper & Thompson(1980). By thorough verification over related corpus data, we conclude that Korean object case marker LUL is not a case assigner but a transitivity marker meaning like {A}`s activity, {O}`s affectedness, kinesis, volition, telic aspect that are internalized in each and every transitive predicate. Besides it, Korean LUL can also occur with NPs and Adjunct that have semantic features like definitive, or specificity.