Erina Murata, Masaki Chujyo, Fujio Toriumi
6 min
Abstract
In Japan, severe rice shortages in 2024 sparked widespread public controversy across both news media and social platforms, culminating in what has been termed the "Reiwa Rice Riot." This study proposes a framework to analyze the temporal dynamics and causal interactions of emotions expressed on X (formerly Twitter) and in news articles, using the "Reiwa Rice Riot" as a case study. While recent studies have shown that emotions mutually influence each other between social and mass media, the patterns and transmission pathways of such emotional shifts remain insufficiently understood. To address this gap, we applied a machine learning-based emotion classification grounded in Plutchik's eight basic emotions to analyze posts from X and domestic news articles. Our findings reveal that emotional shifts and information dissemination on X preceded those in news media. Furthermore, in both media platforms, the fear was initially the most dominant emotion, but over time intersected with hope which ultimately became the prevailing emotion. Our findings suggest that patterns in emotional expressions on social media may serve as a lens for exploring broader social dynamics.
Sam: Yes. This shows emotions don't stay stuck; they evolve as info updates, and X moved quicker. To capture those eight basic feelings—fear, anger, sadness, surprise, anticipation, joy, disgust, trust—they used a setup based on psychologist Robert Plutchik's model. It's like primary colors for emotions: these eight mix and build like red and yellow make orange, giving a fuller picture than just "positive or negative."
Alex: How did they measure those?
Sam: They trained a machine learning tool on Japanese tweets labeled with those emotions at different strengths, so it could score texts reliably. The tool reads posts or articles, estimates probabilities for each of those eight emotions, then averages daily scores into trends. It got about 75% right on test data. This let them plot how fear dominated early, then faded as anticipation rose—smoothed over seven days for steady trends.
Alex: And these emotion trends peaked earlier on X too, matching the post counts?
Sam: Yes. Fear topped out mid-August on X; anticipation overtook it sooner there than in news. Both media calmed by late September. The study suggests X captured raw public reactions—like panic buying reports—faster, pulling news along.
Alex: Does that mean social media caused the news emotions, or just happened first by chance?
Sam: To check direction—not just timing—they measured transfer entropy. Imagine two rivers: one way to tell if upstream flow shapes downstream currents is seeing if knowing the upstream history predicts the downstream better than its own past alone. Here, it showed X emotions reduced uncertainty in forecasting news emotions, especially for fear and anger. Transfer entropy from X to news beat the reverse for all eight emotions overall. Negative ones like fear, anger, disgust flowed strongest from X to news—meaning X surges got reflected in news tone.
Alex: Like X's fear surge made news' next feelings more predictable, proving it led the way?
Sam: Exactly. In August, when crisis hit hardest, X drove most emotions; September saw some flip as news deepened analysis. The evidence points to social media as the starting point for emotional spread in this case, inverting the old mass media lead.
Alex: You've got the trends and the direction measure. But since it's just X, not other apps, does that limit seeing the full picture?
Sam: The paper notes that—sticking to one platform risks missing broader chatter on places like Instagram. Transfer entropy spots info flow, not airtight cause, especially if outside events like government announcements hit both simultaneously. Still, for this riot, it clearly showed X precedence.
Alex: Well, it suggests monitoring social emotions could spot crisis waves early, before news amplifies them. A meaningful shift in how we think about media influence. Thanks, Sam—that's a clear window into how feelings drive crises now. Listeners, check the show notes for the full paper.
Sam: Indeed. The study positions social media as a leading indicator of emotional dynamics, with tools to quantify it. Useful for understanding public discourse in volatile times. Thanks for listening to ResearchPod.