Abstract
This paper presents a mixed-method approach to analyzing news media, combining quantitative linguistic metrics with qualitative discourse frameworks. We first extract linguistic features such as quotations, readability levels, and lexical richness, then perform named entity recognition and topic modeling. To add depth, we apply Fairclough's model of critical discourse analysis-highlighting social and cultural contexts-together with Goffman's frame analysis of social behavior, enabling a systematic comparison of narrative strategies and community engagement. We provide new evidence on how local and national papers diverge not only in content, but also in temporal framing, participatory practices, and the construction of proximity. We validate this pipeline by comparing local and national newspapers in Romandy, Switzerland, a media landscape where local press agencies currently face intense economic pressures and widespread layoffs. Our findings reveal notable divergences in narrative construction and audience engagement: local outlets focus on fostering a sense of community and direct connection with readers, whereas national outlets employ more wide-ranging, sophisticated storytelling to appeal to a broader audience. By synthesizing quantitative and qualitative insights, our study moves beyond descriptive comparison to show how distinct editorial logics shape identity and democratic participation in a transforming media landscape. The study's integrated analytical framework underscores the importance of a comprehensive, multi-level perspective on media strategies and audience dynamics, particularly in an era of rapid editorial and economic transformations.