The Transmission of Nostalgia, Threat and Unity in Populist Communication: Using AI to Analyze Erdogan and Supporters’ Speech

How does Erdogan’s populist rhetoric resonate with his supporters? Our study dives into his use of nostalgia, threat, and unity to rally his base. We found that while Erdogan and his followers align on “threat” messaging—framing enemies as existential dangers—supporters amplify this far more than his calls for unity. Nostalgia for Turkey’s Ottoman past adds a powerful emotional pull, fueling his vision of a revived Islamic identity. These findings reveal how populist language not only connects leaders to their base but also reshapes grassroots narratives, driving division and loyalty in equal measure.

By Matthew J. Belanger*, Ana-Maria Bliuc, John Betts** & Ihsan Yilmaz 

Populist movements have reshaped politics worldwide, and language lies at their heart. This study explores how rhetoric may drive these movements by focusing on Recep Tayyip Erdogan, Turkey’s President, and examining three key strategies in his communication style. Crucially, it also looks at how these strategies echo in the voices of his supporters.

Populism is often defined as a specific political communication style that emphasizes proximity to the “people,” takes an anti-establishment stance, and stresses the homogeneity of the people by excluding certain segments or subgroups of the public (Jagers & Walgrave, 2007). Populist communication is typically polarizing (Bliuc et al, 2024) and juxtaposes the virtuous “people” against a “corrupt elite,” portraying the people as the sole legitimate source of political power (Bonikowski & Gidron, 2016). It combines key messages (content) with stylistic devices (form), often using social media platforms rather than traditional media (Engesser et al., 2017; Ernst et al., 2019).

Moreover, populism is often used as a strategic tool by political challengers, particularly those distant from the center of power, to mobilize support by positioning themselves as outsiders (Bonikowski & Gidron, 2016). Populist communication styles can be understood through the dimensions of identity construction, rhetorical style, and media relationship, and they can influence power relations and political actors of both right and left (Block & Negrine, 2017).

Erdogan has long been established as a populist leader. His communication during critical national events from 2013 to 2023 (for example, national and local elections, the 2016 coup attempt, or the 2017 constitutional referendum) provides an opportunity to understand what rhetorical constructions are present in his discourse and how they can be transmitted and reflected among the population.

Nostalgia has been well-documented in populist rhetoric and typically idealizes collective memories from a supposed golden age (Kenny, 2017; Karakaya, 2018; Elçi, 2021; Menke & Wulf, 2021). Threat, or division, is a further rhetorical device that describes external or internal dangers to create solidarity among supporters as well as a sense of urgency, with xenophobia and racism typically supporting such developments (Rydgren, 2003).

This paper investigates how such rhetorical devices manifest in the rhetoric of both Erdogan and his supporters. Using a combination of thematic analysis and machine learning, we analyze speeches from Erdogan alongside interview data from his grassroots supporters. This study uniquely addresses both the supply of political narratives (the leader’s messaging) and the demand for these narratives (the grassroots reception and reproduction). This allows for a direct comparison of how Erdogan’s language resonates with and influences his supporters.

Erdogan’s Populism

Erdogan’s leadership has reshaped Turkey’s political landscape through a distinctive form of populism often referred to as “civilizational populism.” As the leader of the ruling Justice and Development Party (AKP), Erdogan has framed political conflicts as existential struggles between civilizational identities, casting himself and his party as defenders of Islamic values against the perceived encroachments of secular and Western influences (Yilmaz & Morieson, 2022).

Erdogan’s populist narrative constructs a stark dichotomy between the “righteous people” and the “corrupt elite.” The latter, often characterized as Westernized secularists, are portrayed as detached from Turkey’s authentic Islamic identity (Yilmaz, 2021). These narrative leverages historical grievances, particularly the exclusion of religious conservatives during the Kemalist era, to rally support (Morieson et al., 2024). By drawing upon Turkey’s Ottoman-Islamic heritage, Erdogan positions himself within a broader narrative of resistance against Western hegemony and secular modernity, aligning his leadership with the defense of an Islamic civilization (Yilmaz & Morieson, 2023; Morieson et al., 2024).

Central to Erdogan’s rhetoric is the portrayal of Turkey as besieged by both external adversaries and internal subversives. He often attributes domestic challenges to foreign conspiracies and malign influences, framing himself and his administration as the bulwark against these existential threats (Yilmaz & Shipoli, 2021; Yilmaz et al., 2021). This narrative fosters a siege mentality, consolidating support through fear and the promise of protection (Tas, 2020; Yilmaz & Morieson, 2023).

Following the failed 2016 coup attempt, Erdogan framed the incident as a plot by foreign and domestic enemies, thereby justifying widespread purges within the state and society (Yilmaz, 2021). He depicted the AKP as the ultimate safeguard of Turkey’s sovereignty and Islamic values, branding critics and opponents as traitors and adversaries of both the state and religion (Tas, 2020). He emphasizes national unity, urging all segments of society to come together under his leadership to strengthen Turkey’s cohesion. He calls for solidarity against perceived enemies, emphasizing the need to transcend ethnic, sectarian, and ideological divides to protect the “true people” of Turkey led by himself. By framing himself as the sole political force capable of safeguarding this unity, Erdogan consolidates his support base while marginalizing dissenters as disruptors of national harmony (Yilmaz & Bashirov, 2018; Yilmaz & Morieson, 2023).

Erdogan frequently evokes a romanticized restorative nostalgic vision of the Ottoman Empire, portraying it as a pinnacle of Islamic grandeur and cultural richness. This deliberate invocation of the past serves to cultivate a collective yearning among supporters for a return to Turkey’s former glory (Karakaya, 2018; Elçi, 2021; Yilmaz, 2021). By aligning his leadership with this illustrious heritage, Erdogan crafts a narrative suggesting his governance continues the Ottoman legacy, aiming to reestablish Turkey’s prominence on the global stage (Yilmaz & Bashirov, 2018; Yilmaz & Morieson, 2022).

The Study – What We Did

Our study forms part of an Australian Research Council Discovery Grant project, led by Ihsan Yilmaz and Ana-Maria Bliuc. As a component of the broader project, Erdogan’s speeches from 10 significant national events between 2013 and 2023 were systematically collected. Additionally, 52 of his supporters in Istanbul were interviewed using semi-structured format. The interviews were transcribed and translated into English. The data gathered from Erdogan’s speeches and interviews forms the foundation of our study.

The language analysis was conducted in two primary stages. An inductive thematic analysis was first performed to manually identify examples of two a priori rhetorical devices: nostalgia and threat. These two themes were then used as inputs into a seeded sequential Latent Dirichlet Allocation (LDA) model (Watanabe & Baturo, 2024) which was then used to identify language (words) associated with themes in Erdogan’s speeches. This process also revealed a third residual rhetorical device present in his discourse: unity.

Using a dictionary of thematic labels (terms related to each rhetorical device) identified using LDA, zero-shot text classification (Alcoforado et al., 2022) was then used to identify the presence of these key rhetorical sentiments in Erdogan’s speeches and his supporters’ discourse. The analysis used the xlm-roberta-large-xnli classification model (Hugging Face, 2024; Conneau, 2018) to categorize sentences into the predefined themes: Unity, Nostalgia, Threat, or None, depending on their thematic alignment. A confidence threshold of 0.85 was set to ensure classification accuracy whereby sentences above this threshold were deemed to contain these themes. The comparative analysis then identified the thematic alignment between the leader’s rhetoric and his supporters’ discourse, highlighting similarities and differences in the prevalence of the three rhetorical themes.

Results & Implications

This research quantitatively evaluates the rhetorical alignment and divergence between a populist leader and his supporters. The findings reveal that Unity (8735 sentences; 8.41%) and Threat (8500 sentences; 8.19%) were the most prevalent rhetorical devices in Erdogan’s speeches, with Nostalgia being comparatively less prominent (506 sentences; 0.49%).

Examining the sentences identified as Nostalgia, Unity, and Threat over time, it appears that the proportions of all three remained relatively consistent over time, apart from 2016 which had the highest proportion of threat discourse (i.e., during the 2016 coup).  See Figure 1 below.

Similarly, the interview data from his supporters shows Threat (1563 sentences; 5.51%) as the dominant rhetorical device, followed by Unity (700 sentences; 2.47%) and Nostalgia (52 sentences; 0.18%). These results indicate both alignment and divergence between Erdogan and his supporters, with alignment in the overall prevalence of Threat and Nostalgia, but a notable divergence in the frequency of Unity rhetoric.

The findings suggest that supporters may either amplify threat rhetoric while deprioritizing or underutilizing unity-based messaging. This could reflect a selective resonance with certain aspects of Erdogan’s rhetoric. Threat amplification (i.e., conflict-oriented narratives) may signal the emotional or strategic appeal of threat rhetoric within populist movements, while less emphasis on unity could indicate that it has limited perceived relevance or applicability in grassroots contexts.

This combination of alignment and divergence demonstrates a bit of a dichotomy with respect to rhetorical transmission. On one hand, the alignment demonstrates how populist rhetoric enters supporters’ discourse, and embeds themes of exclusion, moral division, and collective identity into their language. But, on the other hand, the divergence shows that supporters may prioritize certain themes over others based on their own interpretations or circumstances. It could also be the case that unity messaging does not carry the same level of emotional urgency that threat-based rhetoric does. While unity messaging is often framed as positive and affirming, it can lack the intensity that divisive and threatening language provides. Threat rhetoric, on the other hand, inspires more emotional responses such as fear and anger, which are more likely to provoke immediate and powerful reactions. This emotional charge may make threat-oriented messages more compelling and memorable, especially in the context of populist movements where a sense of crisis or danger is central to the narrative.

Through representations of nostalgia, threat, and unity, Erdogan builds a collective identity that contrasts many virtuous “people” with corrupt elites and this method reflects the key themes of populist communication. Our study shows that these themes are not only present in Erdogan’s speeches but are also, somewhat proportionally, internalized and reproduced in the language of his supporters. The language used by political leaders can be viewed as a tool for identity formation, as it can become embedded into the daily speech and thoughts of their followers, as demonstrated within this study. This promotes feelings of solidarity, legitimizes political power, and motivates action.

Overall, our findings show that political leaders can sustain influence over their supporters through carefully crafted rhetorical constructions that resonate with their audience’s values, emotions, and social identities. While far-right populist leaders may claim to promote unity and cohesion, their strategies often involve exploiting social divisions and consolidating national identity in ways that can ultimately reduce social trust and increase tensions between groups (Jay et al., 2019). For example, populism‘s rise is closely linked to xenophobia and political violence through mechanisms of power consolidation, symbolic violence, and socio-political grievances. On top of this, nostalgic constructions create a narrative that contrasts a glorified past with a troubled present, which can further cause unrest. Understanding these dynamics is essential for addressing the challenges posed by populist movements and mitigating their impact on society.

Funding: This work was supported by the Australian Research Council [ARC] under Discovery Grant [DP220100829], Religious Populism, Emotions and Political Mobilisation and ARC [DP230100257] Civilisationist Mobilisation, Digital Technologies and Social Cohesion and Gerda Henkel Foundation, AZ 01/TG/21, Emerging Digital Technologies and the Future of Democracy in the Muslim World.

(*) Dr Matthew J. Belanger is a Lecturer in the Department of Sociology, Social Policy, and Criminology at University of Stirling, United Kingdom. He serves on the research advisory board at the Recovery Outcomes Institute and is on the board of directors at Recovery Scotland. He has a BSc in Kinesiology (2017) from the University of Massachusetts – Amherst, an MSc in Brain Sciences (2019) from the University of Glasgow, and a PhD in Addiction Psychology/Data Science (2024) from the University of Dundee, where he studied biopsychosocial factors influencing addiction recovery. Previously, he worked as a research scientist in the Department of Psychiatry and Psychotherapy at Universitaetsklinikum Carl Gustav Carus in Dresden Germany, undertaking research concerning environmental influences on behaviour. Beyond addiction recovery, Belanger’s interdisciplinary research also heavily involves the application of machine learning in sociological and political contexts.

(**) Dr. John Betts is a Senior Lecturer in the Department of Data Science & AI at the Faculty of Information Technology, Monash University, Australia. He holds a PhD in Operations Research from Monash University, alongside qualifications in statistics, mathematics, and education. His expertise lies in computational modeling, optimization, simulation, and data science, with applications spanning social sciences, medicine, and manufacturing. His research has made significant contributions to areas such as online political behavior, prostate cancer treatment, and Just-in-Time (JIT) manufacturing. Noteworthy collaborations include studies on the influence of online influencers in societal polarization and the effects of local socio-political events on far-right online communities.

References

Alcoforado, A., Ferraz, T. P., Gerber, R., Bustos, E., Oliveira, A. S., Veloso, B. M., … & Costa, A. H. R. (2022) ‘ZeroBERTo: Leveraging zero-shot text classification by topic modeling’, International Conference on Computational Processing of the Portuguese Language, pp. 125-136. Cham: Springer International Publishing.

Bliuc, A-M., Betts, J. M. & Yilmaz, I. (2024) ‘Agent-Based Simulation and Linguistic Analysis of Populist vs. Non-Populist Rhetoric: Insights on Polarization and Cohesion’, European Center for Populism Studies. Available at: www.populismstudies.org/ (Accessed: 6 December 2024).

Block, E. & Negrine, R. (2017) ‘The populist communication style: Toward a critical framework’, International Journal of Communication Systems, 11, pp. 178-197.

Bonikowski, B. & Gidron, N. (2016) ‘The populist style in American politics: Presidential campaign discourse, 1952–1996’, Social Forces, 94(4), pp. 1593-1621.

Conneau, A., Khandelwal, K., Goyal, N., Chaudhary, V., Wenzek, G., Guzmán, F., … & Stoyanov, V. (2020) ‘Unsupervised Cross-lingual Representation Learning at Scale’, Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pp. 8440–8451. Association for Computational Linguistics.

Elçi, E. (2021) ‘Nostalgia as a Political Tool: The Use of Ottoman Imagery in Turkish Populism‘, Middle East Critique, 30(1), pp. 47–62.

Engesser, S., Ernst, N., Esser, F. & Büchel, F. (2017) ‘Populism and social media: How politicians spread a fragmented ideology’, Information, Communication & Society, 20(8), pp. 1109-1126.

Ernst, N., Engesser, S., Büchel, F., Blassnig, S. & Esser, F. (2017) ‘Extreme parties and populism: An analysis of Facebook and Twitter across six countries’, Information, Communication & Society, 20(9), pp. 1347-1364.

Jagers, J. & Walgrave, S. (2007) ‘Populism as political communication style: An empirical study of political parties’ discourse in Belgium’, European Journal of Political Research, 46(3), pp. 319-345.

Jay, S., Batruch, A., Jetten, J., McGarty, C. & Muldoon, O. T. (2019) ‘Economic inequality and the rise of far‐right populism: A social psychological analysis’, Journal of Community & Applied Social Psychology, 29(5), pp. 418-428.

Karakaya, Y. (2018) ‘The Conquest of Hearts: The Central Role of Ottoman Nostalgia within Contemporary Turkish Populism‘, Middle Eastern Studies, 54(5), pp. 790–813.

Kenny, M. (2017) ‘Back to the Populist Future? Understanding Retrogressive Politics in the United States and Europe’, American Affairs, 1(2), pp. 58–72.

Menke, M. & Wulf, T. (2021) ‘The dark side of inspirational pasts: An investigation of nostalgia in right-wing populist communication’, Media and Communication, 9(2), pp. 237-249.

Morieson, Nicholas; Yilmaz, Ihsan & Kenes, Bulent. (2024). “From National to Manufactured: The Evolution of the AKP’s Victimhood Narratives.” Populism & Politics (P&P). European Center for Populism Studies (ECPS). September 6, 2024. https://doi.org/10.55271/pp0040        

Rydgren, J. (2003) ‘Meso-level reasons for racism and xenophobia: Some converging and diverging effects of radical right populism in France and Sweden’, European Journal of Social Theory, 6(1), pp. 45-68.

Tas, H. (2020). “The chronopolitics of Erdogan’s populism in Turkey.” International Political Science Review, 41(4), 632–646. https://doi.org/10.1177/0192512119885444

Watanabe, K. & Baturo, A. (2024) ‘Seeded sequential LDA: A semi-supervised algorithm for topic-specific analysis of sentences’, Social Science Computer Review, 42(1), pp. 224-248.

Yilmaz, I. & Morieson, N. (2023). “Civilizational Populism: Definition, Literature, Theory, and Practice.” In: Religions and the Global Rise of Civilizational Populism. (pp. 1-22). Springer. https://doi.org/10.1007/978-981-99-4262-6_2

Yilmaz, I., & Bashirov, G. (2018). “The AKP after 15 years: Emergence of Erdoganism in Turkey.” Third World Quarterly, 39(9), 1812–1830. https://doi.org/10.1080/01436597.2018.1447371

Yilmaz, I., & Morieson, N. (2022). “Civilizational Populism in Domestic and Foreign Policy: The Case of Turkey.” Religions, 14(5), 631. https://doi.org/10.3390/rel14050631

Yilmaz, I., & Shipoli, E. (2021). Use of past collective traumas, fear and conspiracy theories for securitization of the opposition and authoritarianisation: the Turkish case. Democratization, 29(2), 320–336. https://doi.org/10.1080/13510347.2021.1953992

Yilmaz, I., Shipoli, E., & Demir, M. (2021). Authoritarian resilience through securitization: an Islamist populist party’s co-optation of a secularist far-right party. Democratization, 28(6), 1115–1132. https://doi.org/10.1080/13510347.2021.1891412

Yilmaz, I. (2021). Creating the Desired Citizen: Ideology, State and Islam in Turkey. Cambridge: Cambridge University Press.