HPSG 기반 한국어 문생성과 STYLE 자질

송상헌 경희대학교 언어정보연구소 2009 언어연구 Vol.26 No.3

HPSG (Head-driven Phrase Structure Grammar), a typical unification-based grammar, has proved itself to be a solid foundation for computational implementation, in terms of both parsing and generation. In particular, generation holds a significant position in recent grammar engineering using HPSG. This paper aims to try to generate Korean sentences on the basis of the Korean Resource Grammar (KRG), which is a computational grammar constructed for last several years within the HPSG framework using LKB (Linguistic Knowledge Builder). This paper, to begin with, gives an overall explanation of natural language generation, and then takes two major stages of generation into account; surface realizer and discourse planner. Building upon the fundamentals, this paper proposes how to generate Korean sentences from logical forms (Minimal Recursion Semantics, MRS), and how to choose the results in accordance with stylish properties (STYLE). The whole process has been implemented into computational systems as well, in order to check out its feasibility. The parsing module that this research employs is the PET parser, and the generation module is the LKB system. (Univ. of Washington)

An HPSG based OT Syntax: English Interrogative Constructions

( In Cheol Choi ) 경희대학교 언어연구소 2006 언어연구 Vol.23 No.2

The aim of this article is to examine whether the basic idea and constraints of HPSG framework can coalesce with OT syntax to get over some theoretical and empirical problems that may be confronted by the monotonic constraint based HPSG. Here, I show that the HPSG based OT syntax, as an alternative to the default inheritance system in current HPSG research, well explains inversion, do-support and wh-fronting exhibited by interrogative constructions. In the proposal, those phenomena are explained by the ranked, violable constraints together with HPSG attribute and value system. By doing this, we can obtain the generalization that accounts for the typological characteristics of the language and the explanatory tool that explicitly shows how each universal constraint interacts. (Kyungpook National University)

한국어 문법체계에서 피동접사와 피동사의 처리에 대하여

서민정(Seo, Min-jeong) 부산대학교 인문학연구소 2019 코기토 Vol.- No.87

한국어 문법에서 피동문은 능동문을 전제로 설명된다. 그래서 피동사가 피동문을 형성하는 과정에서도 주로 능동사나 능동문과 관련지어 논의해 왔다. 그러다 보니 피동사를 형성하는 피동접사에 대한 설명에서도 능동사와 관련해서 생각할 수밖에 없고 그러한 맥락에서 피동접사는 능동문을 피동문으로 변형하는 주요한 기제로 다루어지기도 했다. 그에 따라 피동접사는 그것의 통사 기능이 부각되어 설명되기도 했다. 그러나 이 연구에서는 공시적 차원에서 피동접사가 더 이상 생산적인 접사가 아니고, 능동사의 의미와 피동사의 의미가 정확히 일치하지 않는다는 점 등을 근거로 피동접사의 피동사 형성 과정이 통사론적 절차가 아니라 어휘부에서 어휘 생성의 과정으로 이해해야 함을 고찰하였다. 알려져 있다시피 피동 접사에 대한 논의는 오래되었고 다양한 접근들이 시도되었다. 이 연구에서는 먼저 한국어 접사들과 어근의 결합에 대해 파생인가 굴절인가 아니면 통사적 파생인가라고 하는 기존 논의들에 대해 검토하였다. 그리하여 피동 접사가 어휘부에서 이루어지는 문법현상임을 설명하고자 하였다. 그래서 피동접사가 ‘구’에 결합한다는 통사론적 관점에서의 논의에 대해 비판적으로 검토하고, 피동접사가 ‘어휘부’에서 새로운 어휘를 생성하는 파생접사임을 고찰하였다. 또한 이러한 특성을 이론 내에서 설명하기 위해 자질과 제약에 기초한 문법이론인 HPSG의 설명방식을 도입하였다. HPSG의 설명방식은 형태론적 구조와 통사적 기능이 일치하지 않는 한국어 접사의 문법적 특성을 통사론에서 설명하기에 효과적이라고 판단하기 때문이다. 또한 이러한 논의를 통해 형태론적 구조로 이루어져 있으면서 통사론적 기능이 있는 접사들을 문법 체계 내에서 설명할 수 있는 가능성을 제시할 수 있을 것으로 기대한다. In Korean grammar, passive sentences are explained under the premise of active sentences. Thus, the process of forming passive sentences has mainly been discussed in relation to active or active sentences. Therefore, the explanation of passive contact, which forms passive verbs, has to be thought about active participation, and in that context, passive contact has been treated as a major topic of transforming active sentences into passive ones. Therefore, the passive tense was highlighted and explained by its function of syntax. However, it was considered that, at least on a public level, the process of forming passive verbs should be understood in the vocabulary, rather than in the syntactic process, such as the fact that passive contact is no longer a productive one, and the meaning of active speech and the meaning of passive verbs does not match. As is known, the discussion of passive entertainment is old and various approaches have been attempted, and the discussion of this study first examined existing discussions on the combination of Korean speakers and their roots, namely, derivatives, refractions or conventional derivatives. Therefore, this study tried to explain that passive contact is a grammatical phenomenon that takes place in the lexicon. Therefore, the study critically reviewed the discussion from a syntactic perspective that passive particulates are incorporated into "forms", and considered that they are derivative works that produce new words in "wordword phrase". In addition, in order to describe these characteristics within the theory, we would like to introduce an explanation of HPSG, a grammar theory based on qualifications and constraints. This is because the HPSG"s method of explanation determines that the grammatical characteristics of the Korean language, which do not match the morphological structure and syntactic functions, are effective to explain in the general theory.

용언 어미의 통사적 기능에 대한 어휘주의적 해석

서민정,최규수 국제어문학회 2019 국제어문 Vol.0 No.82

Most of the previous studies that have dealt with the Korean verbal-endings are well aware of their syntactic functions. However, there is a difference in the method of explanation. There is an syntactic view (non-lexicalism) in which each verbal-endings is analysed as a single syntaxeme and is analyzed to combine it with other verbal-endings by syntactic actions, such as the head movement. And there is a lexical point of view (lexicalism) in which verbal-endings only exist in lexicon and the lexical formation process is governed by independent principles that differ from the syntactic principle of syntax. The study begins with the premise that grammar, which can explain the syntactic function of the verbal-endings without contradiction within the grammatical system, is grammar that can explain Korean well. So, even if it is possible to explain within a limited scope or simple structure, consider whether it is possible to explain even when applied across the usage language. And this study is not aimed at revealing the fine differences in syntactic and semantic functions of each individual verbal-ending. This study examined the problems and limitations of the explanations of Transformation Grammar and HPSG through “analysis of fusion verbal-endings” and “an analysis of head-complementation structure.” In order to overcome such limitations, the basic principles and methods of explanation of HPSG are introduced. However, it modified and complemented the presenting method of verbal-ending’s grammatical information to suit the characteristics of the Korean. This explained that it is reasonable to analyze Korean verbal-endings in the lexicalism. 한국어 용언 어미를 다루어 온 앞선 연구들은 대부분 어미의 통사적 기능에 대해 잘 인식하고 있다. 그런데 설명방식에서는 차이가 있는데, 어미 각각을 한 개의 통사소로 분석하여 핵어 이동과 같은 통사적 작용에 의해 통사부에서 어간 이나 어미들과 결합한다고 분석하는 통사적 관점(비어휘주의)과 어미들이 어휘 부에서만 존재하며 어휘형성과정은 통사부의 원리와 다른 독립적인 원리에 의해 서 지배된다고 받아들이는 어휘적 관점(어휘주의)으로 구분된다. 이 연구는 어미의 통사적 기능을 문법체계 내에서 모순없이 설명할 수 있는 문법이 한국어를 잘 설명할 수 있는 문법이라는 전제에서 출발한다. 그래서 한정적인 범위 혹은 단순한 구조 안에서는 설명이 가능하다 하더라도 그것이 용언어미 전반에 적용했을 때도 설명 가능한가에 대한 문제를 염두에 두고 논의를 진행하되, 각 개별 어미의 통사적, 의미적 기능의 미세한 차이를 밝히는 것에 목적이 있지는 않다. 이 연구는 먼저 비어휘주의 문법인 변형문법을 도입한 연구와 어휘주의 문법 인 HPSG를 도입한 연구를 중심으로 검토하였다. 먼저 ‘융합형 어말어미의 분 석’과 ‘핵어-보충어 구조 분석’을 통해 변형문법적 연구와 HPSG 연구의 문제점 과 한계를 고찰하였다. 그리고 그러한 한계를 극복하기 위하여 HPSG의 기본 원리와 설명방식은 도입하되, 한국어의 특성에 맞도록 용언 어미의 문법 정보 제시 방법의 수정, 보완을 통해 어휘주의적 관점에서 용언어미를 설명하는 것이 타당함을 설명하고자 하였다.

영어 인상구문에 관한 연구

안종기 대불대학교 1996 論文集 Vol.2 No.1

This paper is to discuss the syntactic and semantic properties of raising construction of English within the framework of Head-driven phrase structure Grammar(henceforth we call it HPSG). We treat mainly the sentences of seem-type verbs and believe type verbs which make raising construction. Both raising construction and equi construction have very similar trees because HPSG does not accept deep structure. But it is natural to say that there are some syntactic and semantic differences between two constructions as the result of the difference of properties. Lexical list and Raising principle can make us distinguish between two constructions. Owing to the lexical list and Raising principle can see some characteristics of raising construction-expletive it as subject of seem-type verbs and as object of believe-type verbs. having no semantic role in the case of passivization of the belief situation, and having no complement omission observed by Jacobson(1990) etc. Furthermore we can also explain some cases not explained by GB framework. Finally HPSG is going forward the lexicalist and constraint-based approach, so I rearranged seem-type verbs and believe-type verbs according to it.

프랑스어 직접의문문에서의 의문사 que의 용법 : HPSG 분석

고길수 한국프랑스학회 2011 한국프랑스학논집 Vol.74 No.-

Nous avons pour but d’examiner l’emploi de que interrogatif dans l’interrogation directe et d’en établir une analyse en HPSG. Le mot que s'utilise devant le verbe, à condition que celui-ci ne soit précédé ni par son sujet ni par une incise, et que son sujet ne soit pas l'objet de la question. Il ne peut apparaître ni après le verbe ni sans celui-ci. Il est aussi impossble de l’utiliser avec la préposition et de le modifier. Il semble donc manquer d’autonomie mais il n'est pas affixe car l'on peut le séparer du verbe avec ne pas, et le combiner avec deux syntagmes verbaux conjoints. Nous le considérons donc comme clitique, et comme étant soumis à des contraintes syntaxiques : le syntagme n’accepte le clitique ni comme la tête ni comme un complément. Le mot que ne peut faire un syntagme ni tout seul ni avec un modifieur. Il reste toujours clitique, ce qui l’empêche d’apparaître après le verbe ou avec une préposition comme complément. Devant le verbe, il est relié au gap, l’élément extrait, alors qu’il ne peut être constituant antéposé car il faut le réaliser avant le sujet. Nous proposons donc un nouveau type de syntagme, qui combine que avec un verbe ou syntagme verbal ayant un sujet non-saturé, et qui le relie au gap enregistré dans le trait non-local GAP. Le syntagme impose d’autres contraintes : il faut que le constituant verbal soit marqué par l’inversion simple, et que son sujet ne soit pas gap.

{이/가}와 {을/를}격교체에 따른 통사-의미구조의 변화와 HPSG 응용

박희문 ( Hee Moon Park ) 한국현대언어학회 2013 언어연구 Vol.29 No.3

The purpose of the study is to investigate the syntactic and semantic changes of the case alternation between Korean nominative particles {-i/-ga} and accusative particles {-ul/-lul} based on the theory of HPSG. The research result reveals that, in addition to the passive and raising verbs, the so-called ‘medio-verbs’ are the major sources of case alternation in Korean. The syntactic features such as [AGT±] and [ALT±], representing agentivity and alternation respectively, were introduced in order to explain the case variation phenomena between nominative and accusative particles. The study also posits the necessity to establish the Theta-role Assignment Rule for COMP-Affix Alternation, characterized by setting up the new “P(rominent)-Theme”, in case when nominative particles {-i/-ga} are instantiated on COMP arguments. (Pai Chai University)

HPSG에 기반한 한국어 문장간 서술대용어의 復元

김정해 가톨릭상지전문대학 1997 논문집 Vol.27 No.1

본 논문에서는 HPSG문법에 기반한 서술대용어를 해결하려고 한다. 사람들은 대화를 할 때 말하고자 하는 주제를 염두에 두고 이야기를 진행한다. 이때 언어의 경제성 논리를 적용하여 같은 말을 중복하여 사용하기보다는 다양한 유형의 대용어를 사용하여 표현하는 경우가 많다. 이러한 언어의 사용은 컴퓨터로 자연 언어를 처리하려 할 때 많은 문제를 유발한다. 따라서 본 논문에서는 대용어의 해결 중에서 특히 문장간에서 야기된 서술대용어의 해결에 중점을 두고자 한다. 이 해결을 위해서는 HPSG에 기반한 파서를 통해 차트 자료 구조상에 보관된 자질 정보들을 이용하였고, 구문적이고 의미적으로 적합한 서술대용어의 선행어구를 추출하여 대용어가 있는 문장을 원래의 문장으로 복원하였다.

Respectively construction in English: A constraint- and construction-based approach

조세연 경희대학교 언어정보연구소 2013 언어연구 Vol.30 No.2

Postal (1998) and Gawron & Kehler (2004) have argued that Respectively Construction (RC) involving filler-gap constructions should be harder to account for under non-transformational approaches. Against their arguments, Chaves (2012), however, has claimed that the RC can be explained in HPSG without any difficulty. In doing so, Chaves proposes an HPSG account of the RC, assuming that sentences with ‘respectively’ will not differ in semantic and syntactic structure from those without it. In this paper, we argue that on the basis of the various properties of the RC, Chaves’ assumption seems to be too strong or inaccurate, though most of his proposals on the RC can be accepted. To support our claim, we provide attested data such as RNRed RC from the corpus or the previous work and seek to give a proper account of RC in HPSG.

Respectively construction in English: A constraint-and construction-based approach

Sae-Youn Cho 경희대학교 언어정보연구소 2013 언어연구 Vol.30 No.2

Postal (1998) and Gawron & Kehler (2004) have argued that Respectively Construction (RC) involving filler-gap constructions should be harder to account for under non-transformational approaches. Against their arguments, Chaves (2012), however, has claimed that the RC can be explained in HPSG without any difficulty. In doing so, Chaves proposes an HPSG account of the RC, assuming that sentences with 'respectively' will not differ in semantic and syntactic structure from those without it. In this paper, we argue that on the basis of the various properties of the RC, Chaves' assumption seems to be too strong or inaccurate, though most of his proposals on the RC can be accepted. To support our claim, we provide attested data such as RNRed RC from the corpus or the previous work and seek to give a proper account of RC in HPSG. (Kangwon National University)