텍스트마이닝을 통해 살펴본 한국의 저출산 담론 변화: 학술논문과 신문 오피니언 기사 비교

초록

This study examines how low fertility discourse in South Korea has been thematically structured and how it has evolved over time in academic articles and newspaper opinion columns (editorials and op-eds), with text mining. The dataset consists of 2,417 academic articles and 4,803 newspaper opinion pieces on low fertility published between 2002 and the first half of 2025. The analysis applies keyword frequency analysis and semantic network analysis (degree and betweenness centrality), and employs Latent Dirichlet Allocation (LDA) to identify and interpret the major themes of the discourse. The results show that academic discourse has evolved from an initial focus on family and childbirth toward a multi-layered thematic structure encompassing childcare and education, labor and welfare, population aging, and regional demographic structures and local governance. An analysis of temporal changes reveals that academic discourse has increasingly reframed low fertility not as an issue of individual fertility choices but as a structural problem embedded in broader social institutions and policy systems. In contrast, newspaper opinion discourse is organized around socially salient issues—such as childcare policy, labor markets, pension reform, immigration, and women’s roles—with thematic emphases shifting relatively rapidly in response to political and economic contexts. These findings indicate that low fertility discourse develops in distinct ways across discursive arenas: while academic discourse accumulates in a cumulative and institutionally oriented manner, media discourse is shaped by immediacy and political salience. By systematically comparing the thematic structures and temporal dynamics of these two discourses through topic modeling, this study empirically demonstrates the gap between academic debates and social perceptions of low fertility, and offers implications for future population policy and policy communication strategies.

키워드