상세 보기
미등록 이주민 뉴스의 귀인 프레임과 이슈 근접성이 댓글 감정과 갖는 조건부 연관성: 딥러닝 모델 기반 텍스트 분석
- 시슈덕;
- 이신행
초록
This study examines how news frames (individual vs. structural attribution) and issue proximity (domestic vs. overseas) in coverage of undocumented immigrants are associated with the emotional expression of online comments, drawing on cognitive appraisal theory and intergroup threat theory. We collected 4,515 news articles on undocumented immigrants published on the Naver news platform from January 2015 to December 2024, together with 137,732 reader comments. News frames were classified with a KPF-BERT-based classifier and comment emotions with a KcELECTRAbased classifier, after which we conducted macro-level time-series trend analyses and micro-level cluster-robust logistic regressions. The main findings are as follows. First, individual-attribution frames were concentrated in domestic issues, whereas structural-attribution frames were distributed relatively neutrally with respect to proximity, revealing an asymmetric coupling between individual-attribution narratives and domestic events in Korean news coverage of undocumented immigrants. Second, the emotional profile of comments shifted toward intensified hostile emotions (anger and disgust) and attenuated sympathetic emotions (sadness and compassion) as well as anxiety, with the onset of the COVID-19 pandemic detected as a key inflection point. Third, the main effect of the individual-attribution frame on hostile emotions was not significant; instead, this frame was negatively associated with sadness. Fourth, comments on articles using structural-attribution frames were significantly more likely to express sadness and compassion. Fifth, with respect to the proximity conditioning of frame–emotion associations, the association between the individual-attribution frame and hostile emotions was not strengthened in domestic issues, whereas the positive association between the structural-attribution frame and compassion was stronger in domestic issues. These results indicate that the associations between news frames and comment emotions vary not only with narrative attributes but also with an issue's spatial proximity. In short, the association between structural-attribution framing and the expression of compassion in comments was pronounced in spatially proximate domestic issues.
키워드
- 제목
- 미등록 이주민 뉴스의 귀인 프레임과 이슈 근접성이 댓글 감정과 갖는 조건부 연관성: 딥러닝 모델 기반 텍스트 분석
- 제목 (타언어)
- Conditional Associations of News Framing and Issue Proximity with Comment Emotions Regarding Undocumented Immigrants: A Deep Learning-Based Text Analysis
- 저자
- 시슈덕; 이신행
- 발행일
- 2026-08
- 유형
- Y
- 저널명
- 한국소통학보
- 권
- 25
- 호
- 3
- 페이지
- 25 ~ 155
- 언어
- KOR
- 출판사
- 한국소통학회
- 발행국가
- 대한민국
- 분량
- 131 페이지
- ISSN
- E 2635-6309
P 1598-7752