Media Framing Analysis of Ethiopian Conflict: An Approach Combining MAXQDA and NLP for Low-resource Languages

Adem Chanie Ali, Seid Muhie Yimam

Abstract: This ongoing research employs computer-assisted methods and NLP techniques to analyze media framing of the Ethiopian conflict in Amharic texts, in two phases. The first phase uses qualitative frame analysis with systematic coding, thematic grouping, pattern detection, and visualization via MAXQDA. It investigates how Ethiopian media depict the conflict in Amhara and Oromia, focusing on framing strategies and responsibility attribution. Analyzing 150 Amharic newspaper articles from Addis Zemen (government-affiliated) and Addis Standard (independent) covering the conflict between 2023–2025. The study, grounded in media framing theory, reveals contrasting patterns: Addis Zemen emphasizes peace, responsibility, and demonization, often externalizing blame, while Addis Standard highlights civilian suffering and shared accountability, especially among the government, Fano, and OLA-Shene. Co-occurrence analysis shows connections between responsibility and humanitarian frames, emphasizing their interrelatedness. This demonstrates digital qualitative methods’ effectiveness in complementing traditional framing analysis. Looking ahead, phase two aims to scale this work by developing NLP techniques such as machine learning classifiers, transformer models, and topic modeling on a larger dataset of approximately 5,000 annotated articles. This dataset, already collected, aims to capture a wider spectrum of conflict-related discourse, integrating qualitative insights with automated NLP to enable scalable, semi-automated conflict framing detection for low-resource languages. The project addresses key challenges in low-resource NLP, including limited annotated data, morphological complexity, and the sensitive nature of conflict discourse, highlighting the potential of combining communication research with advanced NLP to improve multilingual media analysis in conflict zones.

Submission Number: 36