Multimodal Discourse Analysis of Political Speeches in Social Media: A Corpus-Based Study of Rhetorical Strategies Across Platforms

Project Abstract

This study employs a multimodal discourse analysis framework to examine how political speeches are crafted and transmitted across social media platforms, focusing on the integration of verbal, visual, and audial modalities and their rhetorical effects on diverse audiences. By constructing a large-scale, cross-platform corpus of speeches from public political figures, campaign events, and user-generated clips collected from Twitter, Facebook, YouTube, Instagram, and TikTok, the research delineates how platform-specific affordances shape message design, audience engagement, and persuasive outcomes. The methodology combines qualitative multimodal analysis with quantitative corpus-assisted techniques, including transcript alignment, lexical profiling, sentiment and stance detection, visual-semantic tagging of key imagery, and prosodic feature extraction from audio tracks. We introduce a systematic coding scheme for multimodal rhetoric that captures ethos, pathos, and logos across modalities, including facial expressions, gesture dynamics, intonation patterns, captioning practices, meme-based augmentations, and intertextual references. The study investigates three core research questions (1) how do political actors adapt rhetoric across platforms to maximize persuasive impact while maintaining message coherence across multimodal channels, (2) what are the differential rhetorical strategies and audience reception patterns manifested in short-form versus long-form videos, and (3) how do multimodal cues interact with cultural and political contexts to influence trust, polarization, and mobilization. Data are analyzed through three analytical strands (a) a discourse-analytic examination of alignment and disalignment between spoken content and accompanying visuals and sounds, (b) a corpus-based comparison of cross-platform linguistic and semiotic features, including metaphor usage, repetition, framing devices, and visual tropes, and (c) an audience-response analysis leveraging engagement metrics, comment sentiment, and network diffusion patterns to infer reception and influence. The expected contributions include a nuanced taxonomy of multimodal rhetorical strategies in political communication on social media, a cross-platform predictive model of message effectiveness, and methodological guidelines for integrating multimodal analysis with large-scale political communication corpora. The study also addresses ethical considerations such as representativeness, privacy, and potential manipulation, offering a framework for responsible analysis and reporting. Findings are anticipated to reveal platform-driven divergence in rhetorical packaging, with short-form content emphasizing rapid cue-based appeals and longitudinal content enabling complex argumentative trajectories, while visual rhetoric (imagery, color psychology, and facial cues) significantly modulates perceived credibility and urgency. The implications extend to media literacy, policy design, and platform governance, underscoring how multimodal discourse shapes political deliberation in the digital public sphere.

Project Overview

What This Project Is About

---

This project looks at how political speeches are delivered in social media and what makes their messages persuasive. It combines text (written words), images, and sounds to study how different platforms shape the way politicians communicate and how audiences respond.

---

---

The Problem It Addresses

---

People often share political messages across platforms, but each platform can amplify or change the meaning in unique ways. There is a gap in understanding how multimodal cues (words, visuals, and sounds) work together to influence opinions, and how this varies across platforms like Twitter, Facebook, and YouTube.

---

---

Objectives of the Project

---

1. Identify common rhetorical strategies used in political speeches across platforms.
2. Analyze how images, videos, and text interact to shape meaning.
3. Compare how audiences engage with speeches on different platforms.
4. Develop a simple, shareable coding scheme to classify multimodal features.
5. Provide practical insights for media literacy and policy discussions.

---

---

What You Will Do Step by Step

---

1. Formulate a clear research question and select platforms to study.
2. Collect a sample of speeches and accompanying media from each platform.
3. Annotate data for key multimodal features (speech content, visuals, and sound cues).
4. Analyze how these features combine to convey messages and influence perception.
5. Compare patterns across platforms and note differences.
6. Summarize findings and discuss implications for voters and policymakers.

---

---

Expected Outcome

A concise understanding of how multimodal elements shape political messaging online, with practical takeaways for media users, educators, and lawmakers on improving critical consumption of online political content.