A comparative study of emotional narratives in Chinese science fiction: exploring the gender perspective

While Mary Shelley's *Frankenstein* stands as a notable exception, science fiction has historically been dominated by male authors. This imbalance began to shift with the New Wave movement of the 1960s and 70s, and more recently, in China, with the emergence of numerous anthologies showcasing female science fiction writers. A prevailing belief in literary studies suggests that women's inherent sensibility allows them to create narratives with richer emotional depth. This study challenges this assumption by quantitatively analyzing emotional narratives in contemporary Chinese science fiction, using affective computing to systematically evaluate emotional arc, richness, and twistiness in works by both male and female authors.

Existing research offers conflicting perspectives. Some studies, particularly those focusing on autobiographical narratives, suggest differences in emotional expression between male and female authors. Others, using extensive literary corpora, indicate a convergence in emotional expression since 2000. The influence of gender on emotional narratives within specific genres, such as science fiction, requires independent examination. This study addresses this gap by focusing on Chinese science fiction, a genre with a historically lower percentage of female authors.

The study employs affective computing, a field combining computational devices and algorithms to analyze human emotions in text. Two approaches are used: sentiment lexicons and algorithms such as machine learning or deep learning. Given the absence of sentiment-annotated corpora for literary texts, this study utilizes the Chinese Sentiment Vocabulary Ontology Library (CSVOL), a lexicon containing over 20,000 words and expressions annotated with emotion polarity, classification, and intensity. The study constructs two key metrics: 'Richness,' which sums the emotional values across the text to represent the overall emotional intensity, and 'Twistiness,' which measures the fluctuation of emotional intensity across sections of the text. Two corpora are created: one comprising works from 33 female science fiction writers from *She: Classic Works by Chinese Women Science Fiction Writers*, and another with works from 33 male writers, matched for generational distribution and text length. A second corpus, focusing on 12 key writers (6 male, 6 female), with multiple works per author, is also analyzed. Each story is divided into five equal sections, and emotional values are calculated for each section, creating emotional arcs. These arcs are analyzed to compare emotional richness and twistiness between male and female authors, and a similarity metric is introduced to quantify the overlap in emotional narrative styles.

Analysis of emotional arcs reveals no significant differences between male and female science fiction writers in terms of overall emotional expression. While individual authors show varying styles (Figure 1 and 2 illustrate this), the average emotional arcs for both genders show slight variations, but nothing statistically significant. Analysis of richness and twistiness (Figures 3-5, not reproducible here) reveals a wide range of individual styles, but the distributions of male and female writers overlap considerably. A similarity metric, calculated from the areas of the distributions, indicates that the differences within each gender group are as pronounced as those between genders (Figure 6). This suggests that the differences in emotional narratives stem primarily from individual authorial styles rather than gender.

The findings challenge the conventional assumption that female authors inherently express more nuanced emotions in their writing. Three factors might contribute to this: the predominance of male characters in both male and female narratives; the influence of children's literature styles in a significant portion of Chinese science fiction; and the focus on science fiction elements over emotional depth in many works. This study demonstrates the power of large-scale quantitative analysis in challenging preconceived notions in literary studies. The results emphasize the importance of considering individual authorial style over gender in understanding emotional expression in literature.

This study, using affective computing, demonstrates that there is no significant difference in emotional narratives between male and female Chinese science fiction writers. The findings highlight the limitations of generalizations based on limited samples and underscore the value of quantitative analysis in literary studies. Future research could explore other genres or expand the corpus to include a wider range of authors and works, potentially using more sophisticated text segmentation techniques to improve the accuracy of emotional analysis.

The study's limitations include the text truncation method, which may affect context; the limitations of sentiment analysis algorithms in handling stylistic nuances and indirect emotional expression; and the reliance on isolated units of text rather than considering broader narrative context. Future research should address these limitations by developing more sophisticated methodologies for analyzing emotional expression in literature.