Speaker: Ze Yu @Ze_Yu
Affiliation: University of Groningen
Co-authors: Federico Pianzola
Title: Emotions Across Pages: A Comparative Study of Reader Response to Web Novels in Chinese and English on Qidian and WebNovel
Abstract (long version below): This study aims to examine how cross-cultural readers interpret the same stories that are published online. Specifically, it investigates differences in emotional engagement among readers from different cultural backgrounds when interacting with the same web novel published in English on WebNovel and in Chinese on Qidian, focusing on how cultural context shapes reader interpretations.
Working with the Qidian-WebNovel corpus, the study applies topic modeling to classify reader responses and evaluates sentiment scores for reviews of the same topic of the same web novel which quantifies differences in interpretation and engagement among readers across cultures.
Long abstract
The evolution of online reading platforms has transformed reading into a social and interactive activity, offering opportunities for community building and amplifying marginalized voices. Although many empirical studies have examined various aspects of reader response, cross-cultural comparisons—specifically how cultural contexts shape readers’ interpretations—have been largely neglected, with only a few exceptions (Zhang & Laucer, 2017; Hu et al., 2023). Questions about how cultural settings influence the understanding of topics, characters, and plots remain largely underexplored.
A study by Chesnokova et al. (2017) measured the emotions of five hundred language and literature students in response to Poe’s “The Lake” in its translations across Brazilian and Ukrainian editions, showing that reactions to translations often differ significantly from those to the original text. Similarly, Zhang (2022) investigated the interpretation of stories within a psychological situation model framework for Chinese and German children, revealing significant differences in “character judgment,” “plot comprehension,” and “deep inference on characters, plot, and space” between the two cultural groups. Building on this line of inquiry, a comparative study of reader response (Yu & Pianzola, 2024) examined reader response from two reading platforms publishing the same stories in Chinese and English translation. The findings indicate that cross-cultural readers demonstrate both shared and distinct topic engagement with the same story reading, as shown in Table 1. For example, there are some common themes across languages, such as discussions about reading in general, discussions about characterisation and story development, the reading experience, and discussions about cultural references, suggesting that readers often recognise popular cultural elements in the stories and allude tothem in their comments.
Given these shared themes (Yu & Pianzola, 2024) and the fact that when cross-cultural reading happens, the generation of meaning in literary comprehension is first determined by inherent thinking models and relevant cultural characteristics in recipients’ minds (Zhang, 2022). It is worth exploring whether readers engaging with the same discussion topics also express similar emotional responses or whether their interpretations diverge. We believe that understanding this dynamic can provide deeper insights about how cultural or language shapes the way of comprehending and interpreting literary works.
The hypothesis for this study focuses on whether readers of Chinese and English versions of the same stories will exhibit different emotional interpretations in terms of characters, plot, and space. To test this hypothesis, the research draws on a dataset of 110 stories in Chinese and English and over 2 million reviews and replies from two digital social reading platforms, Qidian and WebNovel. The stories are being first published on Qidian in Chinese and later being translated into English and published to WebNovel due to its popularity among fans. Demographically, the readership on Qidian is mainly native Chinese speakers or overseas Chinese-using groups, while readership on WebNovel do not necessarily come from the same region (The top 10 location distribution of readers on WebNovel in the dataset, using the location data available on the users profiles) as shown in Table 2. This ensures the diversity of the backgrounds of the readers of this research cross-culturally.
To ensure the balance of the data, we will select an equal number of books across different genres, focusing on stories with relatively similar ratings and review counts. Methodologically, We plan to employ a newly developed topic model, KeyNMF (Kristensen-McLachlan et al., 2024), which has demonstrated improved performance in processing Chinese characters than BERT. Additionally, we will use a sentiment analysis model (F1-score 0.94) to calculate sentiment scores for reviews (Sharma & Pianzola, 2023). Since the sentiment model was primarily trained on an English dataset, we expanded its training set by incorporating a dataset of social media posts (Weibo) dataset to refine its performance with Chinese text. This approach allows for more effective categorization of reviews and calculation of sentiment scores, as well as better interpret the impact of cross-cultural reading on reader response.
The expected contribution of this study is to bridge the gap between cross-cultural analysis and reader reception studies. By identifying the sentiments in the reviews across different languages that refer to the same topic or content, we aim to gain a deeper understanding of how cross-cultural readers interpret content and in what aspects the difference will demonstrate. We acknowledge that there are certain limitations in our approach for example the translation which may inevitably alter the perception of the stories to some extent. Furthermore, the location of the reader may not fully represent their cultural background. However, this study provides an important opportunity to observe how readers from diverse cultural backgrounds perceive and interact with stories. Ultimately, it can serve as a foundational step for future investigations into cross-cultural reader response analysis.