AI Reveals Emotional Trajectories of Floral Imagery in Tang and Song Dynasties

Scientists at Beijing Normal University are using artificial intelligence to quantify the emotional trajectories of cultural symbols in Chinese history, specifically focusing on the shift from the Tang Dynasty's (618-907) love for peonies to the Song Dynasty's (960-1279) preference for plum blossoms. These floral imagery changes in poetry and visual art reflect broader shifts in societal emotions and aesthetics during these periods. Traditionally, humanities scholars have studied literature and visual art separately, often lacking quantitative methods to analyze emotional evolution. Recently, a team led by Tian Ge Zhou, head of the Art and Technology program at Beijing Normal University's Future Design Academy, leveraged models like BERT, which is based on the Transformer architecture, to conduct a cross-media analysis. These models excel at "distant reading" and can effectively process large volumes of classical texts to extract subtle emotional nuances. Their research, detailed in a paper available on arXiv, reveals that peony imagery in Tang poetry predominantly evokes feelings of joy and anger. Joy reflects the opulence and confidence of the early Tang period, while anger emerges from the political turmoil and social criticism prevalent in the later Tang era. In contrast, plum blossoms in Song poetry and paintings convey emotions of sorrow and fear, aligning with the changing circumstances and internal psychological shifts of the scholarly elite from the Northern to Southern Song. The study also shows clear stylistic changes in floral imagery within paintings, from the grand and exuberant styles of the early Tang to the more restrained and introspective styles of the late Tang. During the Song Dynasty, ink-wash plum blossom branches vary distinctly across key historical periods, further supporting the findings. This data-driven approach effectively combines textual emotion recognition with historical and cultural interpretation, breaking through the longstanding limitation of qualitative dominance in humanities research. It provides a practical framework for studying cross-media symbols and integrates AI into the fields of literature, history, and philosophy. Zhou, who specializes in the intersection of computer technology and artistic expression, and his colleague Shuai Gong, an expert in traditional visual design, initiated this research after a chance academic discussion about the symbolic roles of flowers in Chinese culture. They were intrigued by the stark differences in how peonies and plum blossoms were perceived over time, wondering if these changes reflected not just shifts in taste but deeper emotional currents in history. The team's findings are visually compelling, revealing distinct emotional trends that resonate with the cultural and political climates of the Tang and Song dynasties. This work highlights that data does more than convey information; it captures the nuanced and richly textured destinies and histories of past eras. Looking ahead, the researchers plan to expand their study into a larger, cross-media cultural data platform. This platform aims to bridge the gap between sensory perception and historical and cultural research, leveraging technology to create a more integrated and insightful understanding of the past. By creating historically informed, emotion-guided products for tourism, educational platforms, and serious games, they hope to enhance the spread of cultural knowledge in a manner that is both engaging and deeply resonant. References: https://arxiv.org/pdf/2505.04785