Sentiment Analysis of Tang-Song Poetry and Visual Correspondence
This paper presents a comprehensive study on the interplay between poetic sentiment and visual culture during the Tang (618–907) and Song (960–1279) dynasties of China. By leveraging a BERT-based sentiment analysis framework, the research examines the emotional connotations associated with floral motifs in classical Chinese poetry, specifically focusing on peonies from the Tang period and plum blossoms from the Song period. The study is both ambitious and methodically rigorous, aiming to correlate these literary sentiments with contemporaneous artistic expressions, thereby establishing an empirically-driven framework for cross-media analysis.
The study employs a robust methodology, using a fine-tuned BERT model to process a representative dataset of 100 poems, divided equally between the Tang and Song periods. This method allows for the detection of nuanced shifts in emotional connotations over time. The computational analysis categorizes emotions into seven distinct labels: joy, sadness, anger, fear, surprise, disgust, and neutral. Each of these categories is contextually relevant to the socio-political landscapes of the respective dynasties. For instance, joy is correlated with the celebratory themes predominant in early Tang peony poetry, whereas sadness and fear are more prevalent in the Song era, reflecting the dynasty's political upheaval and cultural introspection.
The results reveal significant temporal shifts in emotional registers between the two periods. During the Tang dynasty, peony imagery predominantly evoked feelings of joy and occasionally anger, mirroring the grandeur and political dynamics of the era. Visual correspondence is found in Tang art, where lush peony depictions convey imperial splendor. The Song dynasty, conversely, expresses heightened tones of sadness and fear through plum imagery, symbolizing resilience amidst socio-political strife. This emotional shift is paralleled in Song art, where the minimalistic portrayal of plum blossoms in ink reflects the emotional sobriety found in the poetry.
The implication of these findings is profound, suggesting a complex interplay between literary sentiment and visual culture that transcends mere aesthetic expression. The results indicate that poetic sentiment not only mirrors but anticipates broader aesthetic trends. Furthermore, the study posits that cultural transmission during these periods was not a simple top-down process but involved a dynamic interplay of literary, artistic, and societal values.
However, the paper also acknowledges certain limitations, such as the fragmentary nature of surviving visual culture and the challenges involved in accurately interpreting classical Chinese metaphors through computational methods. Despite these constraints, the consistency of cross-media correlations strengthens the credibility of the research framework. The study provides a replicable model for future interdisciplinary research, demonstrating how computational methods can augment traditional humanistic inquiry.
Looking forward, this research opens several potential avenues for further exploration. The framework could be adapted to analyze other motifs or applied to different historical contexts. Additionally, the integration of quantitative text analysis with art historical data presents opportunities to further elucidate how aesthetic values have evolved across eras and societies.
In conclusion, this paper enriches our understanding of Tang-Song cultural dynamics by exploring the reciprocal influences between poetic sentiment and artistic representation. It underscores the potential of AI-assisted methods in cultural heritage analysis and offers a model for future studies aiming to investigate the multifaceted nature of historical cultural expressions.
