Yunus Emre Tapan

Investigation of elite opinion on Turkish-American relations

Political elites are known to influence public attitudes and behaviors. In particular, elites play a key role in providing cues and information to the public regarding foreign policy issues. This study is the first attempt to measure prevailing opinions and changes in elite opinions on Twitter in the context of Turkish-American (TR-US) relations. Using snowball sampling, we created a list of 1411 English and Turkish-speaking elites, consisting of diplomats, pundits, academicians, and journalists. To measure elite opinion, we implemented latent dirichlet allocation (LDA) topic modeling. For sentiment analysis, we used bidirectional encoder representations from transformers (BERT). We find that Turkish-speaking elites show more volatility in their sentiments than English-speaking elites. Topic modeling shows that there has been sustained anti-Americanism among Turkish elites in the last five years. Kurds are the main source of disagreement between Turkish and English-speaking elites. For example, English-speaking users prioritize the “War on ISIS” over the “Kurdish issue.” Patriotism and Martyrdom(dying for a nation) were significant motives for Turkish-speaking users. The divergent sentiments and focus of elite tweets show the fragility of Turkish-American relations in recent years and the turning events affecting TR-US relations.

Bio: Yunus Emre Tapan is a doctoral student in the Department of Political Science at Northeastern University, specializing in Political Methodology and Comparative Politics. He earned a graduate certificate program in Computational Social Science. He graduated with an MSc in Middle East Studies from Middle East Technical University and a BA in Economics from Bogazici University in Turkey. Before joining Northeastern, he worked as a Lecturer at Kadir Has University and as a researcher at a non-partisan think tank in Ankara, Turkey. He has a strong background in online extremism and radicalization. He employs machine learning, computational text analysis, and network analysis methods to explore digital trace data.