The image of the European Union (EU) in public opinion remains a difficult topic to gauge, notably due to the specificity of the European construction with different interactions between local, national and EU levels. Yet understanding citizens’ sentiment toward European affairs is crucial, especially as citizen involvement in policymaking is regarded as essential. The European Commission set up in the 1970s a unique collection of opinion surveys, the Eurobarometer, to help assessing the opinion of the Europeans. This source has remained largely unchallenged and is therefore the main dataset used in most of the literature on European sentiment. Nonetheless, the Eurobarometer is not without limitations, notably concerning sampling issues in certain countries and methodological challenges linked to the cross-national intelligibility of its questions across the 27 Member States.
This article precisely aims at providing another source of understanding of the sentiment toward the European affairs in France by drawing on information from the French press. Although the written press is neither fully representative of citizens’ opinions nor as immediate as social media, it remains an important factor in shaping public opinion. As such, it provides a valuable complement to opinion surveys for understanding public sentiment.
Using a large language model (LLM), we built a new dataset covering around 400,00 articles related to the European Union published in France between 2005 and 2023 from more than 100 local and national newspapers as well as magazines. Compared to previous studies (see, for instance, Bortoli et al., 2018), our contribution is twofold. First, we enhance the methodology by going beyond traditional dictionary-based approaches and leveraging large language models (LLMs) to improve classification performance—both in accurately identifying EU-related issues and in assessing article tone. Second, our dataset offers substantially broader coverage than most of the existing literature, which typically focuses on only one or two newspapers (see, for example, Papadia et al., 2019).
Our database helps to better understand the relation of French citizens to the European Union. The salience of EU affairs remained relatively stable over the period, around 0.7% of published articles. Our results further show that between 2005 and 2013, French media sentiment toward the EU closely aligned with Eurobarometer surveys. After 2013, however, the two diverged: the media’s portrayal became steadily more positive, eventually surpassing pre-financial-crisis levels, while Eurobarometer results indicated that public opinion remained stable and less favourable. In addition, the findings highlight that public support for the EU is shaped not only by broad political debates but also by tangible, local initiatives such as school cooperation — areas often underrepresented in national media coverage. In the case of the euro area specifically, we find a closer alignment with Eurobarometer surveys. Although the frequency of coverage is considerably lower, both media sentiment and public opinion show a strong rebound after 2013.
By combining a vast new media dataset with modern AI language analysis, this paper demonstrates how machine learning can uncover richer and sometimes different narratives about public attitudes than those captured by traditional surveys alone. In particular, these results illustrate the multifaceted relationship between French citizens and the European Union as reflected through the press.
Keywords: Sentiment Indicator, European Sentiment, Press Text Mining.
Codes JEL : C55, F59