Earticle

본고에서는 지진과 쓰나미 발생시에 지역주민의 재산과 생명을 보호하기 위해서 신속하게 안전한 장소로 피난을 유도하기 위한 중앙정부와 지방자치단체의 피난방재메 뉴얼에 나타난 피난지시문의 언어적 표현의 특징에 대해서 살펴보았다. 또한, 2011년 동일본대지진과 2024년 노토반도 지진 발생시에 이루어진 NHK재해특별방송에서 사 용된 피난지시문의 언어적 표현과의 비교분석을 하였다. 그 결과 사전에 미리 숙지하는 것을 목적으로 하는 본고의 분석자료와 실제 상황발 생시에 사용되는 피난지시문의 언어적 표현은 차이를 보이고 있는 것을 알 수 있다. 피난을 지시하는 언어적 표현으로는 ‘避難しましょう’, ‘避難してください’와 같은 부드러운 어조의 지시표현과 ‘避難する’, ‘車は使わない’같은 강한 어조의 지시표현(명 령, 금지표현)이 혼재되어 사용되고 있었다. 특히, 중앙부처보다는 지방자치단체에서 이러한 경향이 강하게 나타나고 있었다. 또한, 피난지시문의 구성은 2024년 노토반도 지진시의 NHK재해특별방송은 경우 신속하게 대피하는 것을 강조하는 [긴급표현+피난 지시표현]이 주를 이루었다. 그에 비해 본고의 피난지시문은 신속하게 대피하는 것보 다는 안전한 장소로 대피하기 위해서 안전한 장소에 대한 정보를 제공하는 [피난장소+ 피난지시표현]으로 구성되는 피난지시문이 사용되고 있었다.

In the present study, we examined the characteristics of the linguistic expressions of evacuation instructions in the evacuation and disaster prevention manuals of both central and local governments. The disaster prevention manuals are published to promptly guide residents in the events of earthquake or tsunami to evacuate to safe places in order to protect their property and lives. Results showed that there were differences between the linguistic expressions of evacuation instructions used in actual situations and the analysis data of this study, which aims to familiarize people with the evacuation language in advance. In terms of linguistic expressions directing evacuation, both soft–toned directive expressions, such as “避難しましょう”, “避難してくださ い”, and strong–toned directive expressions (orders, prohibitions), such as “避難する”, “車は使 わない”, were used in a mixed manner. In addition, the composition of the evacuation instructions mainly consisted of [emergency expressions + evacuation instruction expressions] that emphasized quick evacuation, as found in the NHK disaster special broadcast for the 2024 Noto Peninsula earthquake. By contrast, for the disaster prevention manual mentioned in the present paper, the evacuation instructions were composed of [evacuation place + evacuation instruction expressions] that provide information on evacuation to a safe place rather than quick evacuation.

Local governments need to continuously enhance their staff’s disaster response capabilities by incorporating the latest knowledge and experience. Recently, after-action reports have been frequently published by local governments to review their disaster responses. Although learning from these reports is effective for improving disaster response capacity, it is not easy for local government officers to spend sufficient time learning from them. In this study, a method to efficiently extract lessons learned from after-action reports on local government disaster response was developed. First, cases of critical situations during emergency responses were collected from after-action reports published after recent disasters. Next, a deep learning model based on a language model (BERT) was developed using the collected cases as training data. The model enables the extraction of new cases similar to the collected ones from after-action reports. As the developed model was able to extract new cases with high accuracy, it may contribute to the efficient learning and analysis of lessons learned from after-action reports. On the other hand, the findings suggested the need to explore the optimal amount of training data required to achieve high accuracy and stable learning performance. After these efforts were started with after-action reports of flood disasters, the reports of earthquake and tsunami disasters were also added as the targets. Finally, a support system was developed to extract lessons from after-action reports using the proposed model and to present accumulated lessons efficiently. This system enables users to explore the latest lessons from recently published after-action reports and enhance their capabilities based on these lessons.

This study examines disaster awareness and preparedness among international students in Japan, with a focus on the role of Japanese language proficiency. A questionnaire survey was conducted with 195 international students in Tochigi Prefecture, followed by chi-square analysis to explore differences between students who use Japanese and those who do not. The findings indicate that a high proportion of international students (over 85%) possess sufficient Japanese language skills for daily communication, while about 15% have limited proficiency. Only around 30% had prior disaster experience in their home countries, highlighting the need for targeted support, particularly for newly arrived students. Universities play a critical role in providing orientation and training to improve students’ understanding of disasters in Japan. The study also reveals that international students tend to evacuate to familiar locations, such as their universities or places where acquaintances are present, rather than officially designated evacuation centers. This suggests the importance of preparing non-designated locations, especially universities, to function effectively as evacuation sites during emergencies. Regarding the relationship between language ability and disaster awareness, students who can speak Japanese demonstrate greater knowledge of evacuation centers. However, Japanese proficiency does not necessarily reduce anxiety or improve overall preparedness. In some areas, such as first aid knowledge and communication with local residents, students with limited Japanese skills showed higher levels of preparedness. These results indicate that disaster preparedness education should be provided to all students, regardless of language ability. Finally, the study highlights limited interaction between international students and local communities. Strengthening these connections is essential, as international students—particularly those with strong Japanese skills—have the potential to contribute as disaster volunteers and community leaders, supporting mutual aid in aging and depopulating regions.

본 연구는 언어네트워크분석을 통해 재난 정신건강 관련 주요 학문분야인 의학, 심리학, 사회복지학 분야에서 재난 연구동향을 살펴보고 이를 비교하는 것을 목적으로 하였다. 이를 위해 1996년부터 2015년까지 출판된 재난 관련 학술 및 학위논문 총 152편(의학 62편, 심리학 48편, 사회복지학 42편) 을 대상으로 UCINET 프로그램을 활용하여 각 분야에서 중요시하고 있는 연구 주제어들을 검토하고 주제어들 간의 유기적인 관계를 분석하였다. 그 결과, 의학과 심리학 분야에서는 ‘비전염성 사회재 난’, ‘피해자’, ‘PTSD’ 등 특정 재난 사건에 대한 피해자들의 정신과적 증상이나 심리적 과정들이 중심 연구 주제어로 확인되었다. 반면, 사회복지학 분야에서는 ‘재난’, ‘사회복지’, ‘관리’, ‘체계’, ‘조직’ 등 일반적인 재난 상황에서의 피해자 지원 관리 체계가 보다 중요한 연구 주제들로 다루어져 왔음을 알 수 있었다. 본 연구 결과를 토대로 향후 정신건강 분야의 재난 연구에 있어 학제 간 연계의 필요성을 제안하였다.

The purpose of this study is to analyze research trends related to disaster in the filed of Korean domestic mental health (i.e., Medicine, Psychology, Social welfare). To do this, 152 articles published from 1996 to 2015 were selected: 62 for Medicine, 48 for Psychology, and 42 for Social welfare. Then, a language network analysis is conducted by using the UNCINET program. The result shows that in Medicine and Psychology, the key words such as “noninfectious social disasters”, “victim”, “PTSD” have the high value of centrality, which indicates that the studies about psychiatric symptoms or psychological mechanism of the victims in specific disaster events are most frequently published in Korean academic journals. In the filed of social welfare, the key words such as “disaster”, “management”, “system”, and “organization” have the high values of centrality, which implies that most studies have strong interests in management systems when general accidents and emergency situations occurred. Based on this result, the direction of further research about the disaster management in mental health fields will be discussed.

인공지능의 눈부신 발전에 따라 여러 분야에서 인공지능 모델을 사용하기 시작했고, 이와 동시에 인공지능에도 보안의 허점이 드리워졌다. 본 논문은 재난 상황에 사용되는 생성형 모델에 데이터 오염과 전이학습을 통해 백도어를 삽입하는 상황을 가정한다. 해당 백도어를 통해 일반적인 입력에 대해서는 정상적인 출력을 내놓지만, 특정 트리거가 포함된 입력에 대해서는 의도적으로 변조된 출력을 내놓도록 하여 사회에 혼란을 야기할 수 있음을 실험적으로 보여주고, 위험성을 강조한다.

With the remarkable advancement of artificial intelligence, AI models have begun to be used across various fields, while simultaneously revealing security vulnerabilities within AI itself. This paper assumes a situation in which a backdoor is inserted into a generative model used in disaster situations through data poisoning and transfer learning. By means of this backdoor, the model produces normal outputs for general inputs, yet deliberately yields manipulated outputs for inputs containing a specific trigger, thereby potentially causing social confusion. We experimentally demonstrate this possibility and emphasize its associated risks.

본 연구는 언어 네트워크 분석을 통해 재난정보와 관련한 국내 연구동향 분석을 목적으로 한다. 이를 위하여 학술연구정보서비스(RISS)를 검색하여 2008년부터 2017년 사이에 발간된 재난정보와 관련한 국내 학위논문 및 학술지논문 312건을 수집하였다. 그리고 논문들의 서지사항을 토대로 통계분석을 실시하였다. 뿐만 아니라 연구논문들의 논문명을 대상으로 키워드를 추출하여 빈도분석 및 언어 네트워크 분석을 실시하였다. 분석 결과, 최근 재난분야에서 빅데이터와 관련한 연구가 급증하였으며, 재난정보 공유 및 활용의 중요성이 증대되고 있다. 또한 재난대응을 위하여 공간정보, 실시간정보, 지리정보 등 다양한 유형의 재난정보가 활용되고 있었다.

This study aims to investigate research trends on disaster safety Information based on the language network analysis methods. To accomplish it, we collected 312 Korean thesis and scholarly articles on disaster safety information published between 2008 and 2017 from RISS (Research Information Sharing Service) site. With the collected data, this study performed the statistical analysis based on bibliographic data. Also, this study performed the analysis of frequency and language network on keyword extracted from titles on the collected scholarly articles and thesis. This study found out that researches recently on Bigdata related to disaster safety information have been rapidly increased. Also, the needs of sharing and utilizing disaster safety information have increased. Also the various types of disaster safety information such as spatial data, real-time information, geographic information has been used for the disaster response.

본 연구는 재난관리 분야에서 법률용어가 아니지만 일반적으로 많이 사용하는 외국어와 외래어의 재난관리 전문용어 (terminology)의 통일화 방안에 목적이 있다. 재난관리에 있어 커뮤니케이션은 매우 중요한 부분으로 용어정의가 명확하지 않 아 의미전달에 문제가 있어왔다. 「재난 및 안전관리 기본법」 34조의4 및 동시행령 제43조의5의 기능별 재난대응활동계획에 13개 협업기능을 위해서는 커뮤니케이션이 필요하다. 이러한 협업기능을 수행하기 위해서는 기능별로 상호간에 정확한 개념 표현의 의사전달이 되어야 한다. 따라서 본 연구에서는 재난관리 전문용어로 사용하는 외국어 및 외래어의 개념과 정의를 살펴 보고 재난관리에 사용하는 전문용어가 바르게 사용되는 표준화를 위한 토대를 만들고자 한다.

With the recent development of big data, disaster management provides access to a lot of information as well as a variety of language services. Natural and man-made disasters have global repercussions, like the current coronavirus pandemic. In view of the increase in disasters and the development of information and communication technologies, it seems opportune to conduct a study on the evolution of language related to disasters. We will first focus on the theories which relate catastrophes and language and we will take a particular interest in the analysis of the linguistic network. This theoretical basis will allow us to approach the way in which the French language to integrate or reject, transform and then shape a series of terms conjunctive to COVID-19. Eleven new uses will then be selected - according to criteria of frequency of use, disciplinary convergence, association with other neologisms and cultural relevance - then analyzed. The results of our research will show that it is possible to teach these new words to French language learners. To this end, the last part of our study will propose a course model structured around five stages: identification of new uses, polysemy of terms, association of ideas, words in networks and narration.

This study, as part of the Language Culture Improvement Movement, examines the language use in disaster broadcasting with the aim of preparing guidelines for appropriate language use in disaster news reporting. The analysis is carried out on news reports on the Sewol ferry disaster between April 10, 2014 and April 30, 2014 by three terrestrial broadcasters(KBS News 9, MBC Newsdesk, SBS 8 News) and three general cable TV channels(ChannelA News, MBN News 8, JTBC News 9). The collected news reports analyzed in terms of the following five criteria: normativeness(ungrammatical or unnatural expressions) accuracy(unfounded or unsourced information, conjectures, exaggeration, or abstruse expressions), decency(sensational or crude language), fairness(emotionally-charged, discriminatory, or biased language), comprehensibility(hard to understand terminology or Chinese and foreign words). The results are as follows: 20.9% of abstract words, 13.9% of obscure Chinese and foreign words, 13.0% of sensational language, 10.3%, of conjecture words, 10.2% of biased language, 7.6% of unfounded or unsourced information, 6.0% of non-normative and 5.9% of exaggerated word. Based on the results, the following measures for improvement are proposed. First disaster broadcasting should use language whose sentence constituents grammatically and semantically agree with each other for the speedy delivery of relevant information. Second, disaster broadcasting should reveal details of the source of information. Third, disaster broadcasting should refrain from using conjectures, and acknowledge their use if unavoidable. Fourth, disaster broadcasting should not exaggerate in order to stress or highlight 'a certain situation'. Fifth, disaster broadcasting should use plain language for accurate delivery of information.

본 연구는 환경재난의 대표적 사례로서 가습기살균제 참사 관련 기사 속에서 개별 행위자들을 확인하고, 이들 간의 연결망을 분석하여 환경재난에 대한 사회복지영역의 개입 가능성을 확인하고자 하는 목적으로 수행되었다. 이를 위해 우리나라 5대 일간지(경향, 동아, 조선, 중앙, 한겨레) 기사를 활용하여 언어네트워크 분석을 실시하였다. 행위자(피해자, 가해자, 매개집단)를 중심으로 연결정도중심성과 매개중심성을 파악해 본 결과, 상위 15개 노드를 기준으로 유의한 차이를 확인할 수 있었다. 피해집단에서는 문제가 공론화되는 시기가, 가해집단에는 법정진행 상황이, 매개집단의 경우 사건을 포괄적으로 포함하는 개념(안전)이 대표적인 노드들이었다. 다만 상위 10개 노드만 비교할 경우 행위자간 유의한 차이를 확인하기 어려웠는데, 이는 다수의 기사들이 문제를 실제로 경험하고 조정하고 있는 행위자의 입장을 심층적으로 확인하고 살펴보기보다는, 발생한 사건 자체를 확인시키는 수준에서 그쳤기 때문인 것으로 판단된다. 다양한 행위자들의 상호작용을 확인할 수 있는 언론의 특징을 고려해볼 때, 환경재난에 대한 언론의 경직된 태도는 대중이 가습기살균제 피해자에 대한 관심을 높이기 어려운 구조인 것이다. 환경재난은 수많은 피해자를 발생시키고, 그로인한 피해는 개인적인 노력으로는 회복이 불가능하다. 이에 본 연구는 사회복지학에서 상대적으로 잘 다루어지지 않았지만, 현대와 같은 위험사회에서 피할 수 없는 환경재난(가습기살균제 참사)에 대한 개입가능성을 확인하고, 동시에 개입을 위한 가장 기본적 전제인 행위자에 대한 이해를 제공하였다는 점에서 의의를 가진다고 할 수 있다.

This study was carried out in order to identity the possibility of intervention by social welfare area to environmental disaster by checking the individual actors from the related articles about humidifier disinfectant disaster, which is a representative example of environmental disaster, by analyzing its network. For this goal, the language network analysis was conducted by utilizing the articles of Korea’s top 5 newspapers (Gyeonghyang, Donga, Chosun, Joongang, Hankyoreh). As a result of analyzing the connection level centrality and mediation centrality focusing on actors (victim, perpetrator, mediation group), the significant different was detected based on top 15 nodes. The representative nodes were the time when the problem was publicized in the victim group, the court proceeding in the assault group, and the concept (safety) that comprehensively contains the events was the representative nodes to the mediating group. However, when comparing top 10 nodes only, it was difficult to identify the significant different between actors, it was judged that’s because many of articles just remain at the level that just confirms the accident itself rather than intensively verifies the situation of actor who experiences and adjusts the problem. Considering the characteristics of press that can identify the interaction of various actors, the media 's inflexible attitude toward environmental disasters is a structure in which the public can’t raise interest in victims of humidifier disinfection. Environmental disaster produces many victims, the damage can’t be restored by the individual effort only. Accordingly this study is meaningful in that it identified the possibility of intervention in an inevitable environmental disaster (humidifier disinfectant disaster) in a dangerous society even if it was not addressed relatively well in social welfare, and at the same time, it provided an understanding about actor, which is the most basic premise.

이 연구는 사람이 비적형 문장(non-fully well-formed sentence)을 처리할 때 보이는 유연성(flexibility)을 언어모델에서도 관찰할 수 있는지를 분석하였다. 언어모델이 사람과 같은 수준의 언어 능력을 갖추었는지 평가하기 위해서는 비적형 문장과 같이 유연성을 요구하는 벤치마크의 도입이 필요하다. 이를 위해, 재난문자를 기반으로 적형문과 비적형문 쌍으로 구성된 542개의 데이터세트를 구축하였다. 언어모델의 평가 방식은 이해와 생성의 두 가지 방식으로 진행되었으며, 각각의 과업에 특화된 언어모델을 선정하였다. 이해 실험은 5종의 인코더 기반 언어모델(KLUE-RoBERTa, KoBERT, KR-BERT, mBERT, KoELECTRA)에 자연어 추론(Natural Language Inference) 과업을 적용하였다. 생성 실험은 디코더 기반 언어모델(ChatGPT-4)이 생성한 전보문에 대해 설문조사를 진행하였다. 실험 결과, 인코더 기반 언어모델은 비적형 문장을 처리하는 과정에서 일관되지 않은 결과를 나타냈고 디코더 기반 언어모델이 작성한 전보문은 사람이 작성한 전보문과 큰 차이를 보이지 않았다.

This study investigates whether Korean language models can flexibly process non-fully well-formed sentences as they do fully well-formed sentences. For this purpose, we constructed a dataset of well-formed and telegraphic sentence pairs using disaster messages. We conducted two experiments utilizing this data: one with an understanding task on five encoder-based models (KLUE-RoBERTa, KoBERT, KR-BERT, mBERT, KoELECTRA) and the other with a generation task on a decoder-based model (ChatGPT-4). The results of the experiments revealed that language models show inconsistent performance in understanding and exhibit illogical patterns in generation, highlighting a gap between human language competence and language model's performance. Based on these findings, we propose the importance of robustness in language model evaluation methods and emphasize incorporating qualitative assessments to better reflect the complexity of human language. This study suggests that there remain significant challenges for language models to more accurately mimic human understanding and generation.

본 연구의 목적은 재난 연구 동향과 재난안전 R&D가 유사한 방향성을 지니고 있는지를 실증적으로 분석하는 것이다. 보다 구체적으로 재난안전 R&D 사업을 내용분석하고, 재난 연구 동향을 연결망 분석 특히 언어네트워크를 활용하여 분석하였다. 2007년부터 2018년까지 총 12년간의 국내 주요 연구논문 및 2018년 재난안전 R&D 사업(17개 부처)을 분석한 결과, 재난 연구 동향과 재난안전 R&D 사업의 연계성이 부족하고 전체적인 방향성이 차이 나는 것을 확인하였다. 따라서 본 연구는 재난안전 R&D가 보다 장기적인 시각에서의 사업 계획과 전략 수립, 포괄적 재난대응을 위한 균형과 배분, 특정 재난 연구 이후의 재난 연쇄성을 고려한 후속 연구사업의 기획을 제언하고자 한다.

The purpose of this study is to analyze empirically whether disaster research trends and disaster safety R & D have similar direction. More specifically, disaster safety R & D projects were contents analyzed and disaster research trends were analyzed using language network analysis. As a result of analyzing major domestic research papers for 12 years from 2007 to 2018 and disaster safety R & D projects (17 ministries) in 2018, It was confirmed that there is a lack of connectivity between disaster research trends and disaster safety R & D projects and that the overall direction is different. Therefore, this study suggests the plan of disaster safety R & D to establish business plan and strategy from a longer-term perspective, balance and allocation for comprehensive disaster response, and planning of follow-up research considering disaster successiveness after specific disaster research.

본연구에서는재난시정보약자인외국인으로서일본인한국어학습자의입장에서한국어재난안전문자의어휘,문법,문말표현을조사하여학습자의이해를저해하는요인을분석한것이다. 한국은코로나팬데믹을계기로정부,지방자치단체로부터실시간으로재난안전문자가전송되고있다. 재난발생시증가하는외국인들의모국어에대응하여필요한정보를동시에번역해서전달하는것은불가능하다.정보전달을위한언어는한국어가효율적이며외국인에게쉽고간결한한국어로전달하는것은최소한의배려이다.그러나한국의재난안전문자는한국인도이해하기어려운표현,비문법적인표현등으로외국인은물론한국인도이를이해하는데어려움을겪고있다.재난안전문자는생명과직결되므로평이한어휘와문장으로필요한정보를간결하게전달하는것이중요하다.한국사회가다문화사회로접어든만큼사회구성원으로서외국인에대한효과적인정보전달방법에대한재고가필요하다. 본연구에서는일본인한국어학습자를대상으로재난안전문자이해도에대한설문조사를실시하여학습자들이어려워하는한국어재난안전문자의어휘리스트,문법표현,지시문의문말표현을제시하였다. 이를토대로한국어교육의관점에서이를분석하고,재난정보문에서일본인학습자들의이해를저해하는요인을제시한후,쉬운언어적관점에서서바이벌교육을제안하였다.일본인에한정된이번연구에이어서앞으로본연구는다양한국가의외국인을대상으로범위를확대하고명확하고간결하며쉬운표현의재난안전문자를작성하기위한기준을마련하는데매진할것이다.

It might be difficult for foreign residents of Korea to respond quickly to disasters. This study looks at safety protocols in relation to language issues in the event of an emergency. This research also focuses on ambiguous amber signals, which confuse Japanese residents in Korea due to difficulties with vocabulary, grammar, and expressions in these texts. Throughout COVID, the local community center and the Korean government did not stop issuing amber warnings. During a pandemic, it would be difficult to simultaneously translate and broadcast emergency notifications with all necessary information in all languages. Developing and using emergency warnings for foreigners in plain language is thus strongly advised. Complicated emergency warnings produced for native speakers might be incomprehensible to foreigners with language barriers. Korea needs to develop these notifications more efficiently since it is now a multi-cultural nation. In order to determine how effectively Japanese (Korean Learners) comprehend notifications, this research conducted a survey. Grammar, sentence construction, and vocabulary lists are all covered in this examination. The findings suggest that communication methods should be taken into account from an easier-to-learn standpoint. We expect that clearer amber notifications for international visitors to Korea will prove extremely helpful.

코로나 19 같은 긴급한 사회적 재난 상황에서 청각장애인은 재난 지리정보 취득과 접근이 제한되고 있어 이동권 뿐아니라 안전권 확보가 어려워지고 있다. 청각장애인에게 필요한 지리정보를 적절하게 제공하기 위해서는 이들의 장애 특성과 의사소통 수단을 이해하는 것이 필요하다. 이와 같은 배경에서 본 연구는 지리학적인 관점에서 청각장애인의 주된 의사소통 방법인 수어의 공간성을 탐색하였다. 수어의 공간성은 수화 표현의 공간성, 수화 대화의 공간성, 지리적 요인에 의한수화 변화로 나누어 진다. 본 연구는 농 문화에 공간적인 요소를 융합한 농 지리학 연구로서 청각장애인의 삶의 질 향상에기여할 수 있다.

In an social disaster situation such as Corona 19, it is becoming difficult for the deaf to secure safety rights as well as mobility rights, as access to and access to disaster geographic information is limited. In order to properly provide the geographic information necessary for the deaf, it is necessary to understand their disability characteristics and means of communication. In this background, this study explored the spatiality of sign language from the geographical point of view. The spatiality of sign language is divided into the spatiality of sign language expression, the spatiality of sign language discourse and sign language change due to geographical factors. This study is Deaf geography that fuses spatial elements with Deaf culture. The results of this study may contribute to improving the quality of life for deaf people.

1930년대 중반 식민지 조선사회는 대공황 여파와 각종 재난에 대한 구제, 구호가 절실한 시기였다. 그 가운데 삼남지방은 1933년, 1934년, 1936년에 심각한 수해를 겪었다. 특히 낙동강 연안의 하류 지역은 대대적인 하천개수공사가 진행되는 과정 중에 1933년과 1934년의 연이은 피해를 입었다. 1935년에는 직접적인 대형 수해가 없었으나, 지역주민들은 앞선 대형 재난의 여파와 국지적 재해들을 경험했고, 이듬해 1936년에는 또 다시 전국적 규모의 수해로 큰 피해를 입었다. 이러한 ‘연속적’ 수해를 겪는 과정에 여러 층위, 다양한 방면의 대응 조치가 이루어졌고, 지역사회와 식민지 조선 전반에 영향을 미쳤다. 이에 본고에서는 1930년대 중반 낙동강 유역의 수해 경과를 구체적으로 살펴보고자 했다. 한편, ‘연속적’ 수해의 경과 과정 속에서 언론도 하나의 대응 주체로 존재했다. 중앙에서 발행된 세 한글 민간지(『동아일보』･『조선일보』･『조선중앙일보』)는 기본적인 기능을 발휘하는 형태로서, 재난 보도와 직접 구제활동 참여하는 방식으로 지역의 대형 재난에 대응했다. 그러나 일본제국과 조선총독부의 식민지적 통치질서가 강하게 작용하는 상황에서 그 활동은 여러 제약을 가지고 있었다. 그리고 다른 한편으로는 당시 언론의 상업주의 강화 경향 속에서 재난을 하나의 계기로 경쟁우위를 획득하려는 경쟁적이며 자본친화적인 면모도 드러나고 있었다.

In the mid-1930s, colonized Korea society was in dire need of relief from the aftermath of the Great Depression and various natural disasters. Three southern provinces (Gyeongsang-do, Jeolla-do, and Chungcheong-do) suffered severe floods in 1933, 1934, and 1936. In particular, the lower coasts of Nakdong River were damaged by two consecutive floods in 1933 and 1934 during the major river maintenance projects. Although there was no major flood in 1935, local residents suffered from the aftermath of the previous major disasters and other minor ones, only to followed by the nationwide flood in 1936. These “successive” disasters resulted in responses from various levels of society, affecting local communities and colonized Korea as a whole. This article covers specifically the progresses of floods in the Nakdong River basin in the mid-1930s. The media also played its role in the course of the “successive” floods. The three national Korean-language private newspapers (The Korea Central Daily News, The Korea Daily News, The Oriental Daily News) responded to large-scale disasters in the region by reporting the progress and participating in direct relief efforts. These activities, however, were heavily intervened the Japanese Government-General of Korea. On the other hand, the increasing commercialism of the media at the time revealed a competitive and capital-friendly side that sought to take advantages through disasters.