Yesterday once more: collective storytelling and public engagement with digital cultural products on the music streaming platform

The study asks how audience emotions and memories evoked by music, expressed through user comments, shape a song’s influence on streaming platforms. Grounded in narrative transportation theory, the authors argue that listeners are transported into autobiographical memories and collectively narrate their experiences in comment sections. Because most cultural products are now disseminated via social platforms, user comments may amplify popularity and perceived quality. The authors conceptualize these comments as collective storytelling that creates a virtual narrative world, potentially influencing engagement (comments, likes) and ratings. Using NetEase Cloud Music as the empirical site, the study examines whether narrative themes and emotional framing in comments are associated with higher user engagement and perceived quality, and whether matching between song content and comment framing further amplifies influence.

The theoretical framework integrates source-level and message-level determinants of cultural product influence with narrative transportation theory. Prior work shows success-breeds-success dynamics, bandwagon cues, and effects of content quality and identity cues on engagement. The paper extends this by focusing on user comments as narrators that contribute to sense-making beyond simple social cues. Narrative transportation suggests that emotional reactions and autobiographical memories can shape attitudes and behaviors. On music platforms, user comments often involve personal stories and emotions that can foster resonance and community. The authors develop three hypotheses: H1, that the amount of narrative themes in aggregated comments positively relates to user engagement and perceived quality; H2, that the strength of emotions in aggregated comments positively relates to user engagement and perceived quality; and H3, that the effect of song content on engagement and perceived quality is moderated by the emotional framing and narrative themes of comments (i.e., matching effects).

Data were collected from NetEase Cloud Music (mainland China) via a Python web crawler. The scraping proceeded by singers, albums, songs, and comments, focusing on Chinese singers (male, female, bands). The dataset includes 7470 popular Chinese singers, 137,546 albums, 478,864 songs, and 1.31 million top-rated comments drawn from 187,678 songs (each song page shows up to 15 top comments sorted by likes). For each comment, text, time, likes, user ID, and total number of song comments were collected; user profiles (e.g., gender, registration date, region, fans, followers, playlists) were also scraped. Ethical safeguards included only public data, de-identification, secure storage, and aggregate reporting. Text preprocessing removed punctuation, numbers, URLs, non-Chinese characters, and stopwords; Chinese word segmentation used Jieba. Dependent variables: user engagement measured as log number of comments per song; perceived quality measured as the platform’s song score (log-transformed). Predictors and controls: number of likes on comments (log) as a control for social endorsement; singer type (male, female, band) and number of albums (log). Emotion framing: sentiment polarity (negative probability) for lyrics and comments using BosonNLP; seven discrete emotions (love, happiness, sadness, anger, fear, disgust, surprise) in lyrics and comments using the Chinese Emotional Vocabulary Ontology. Topics: LDA topic modeling (Gensim) with five topics selected by perplexity for both comments and lyrics. Comment topics included expressing feelings about music, views on films/TV dramas, evaluating song creation, evaluating singer performance, and nostalgia. Lyric topics included pursuit of ideal life, nostalgia, gratitude to family/friends, love, and martial arts themes. Analysis used fixed-effect regression models at the singer level to test main and interaction effects, including comment–lyric matching interactions.

- Source- and platform cues: Compared to bands, male singers were associated with higher engagement (B=0.144, p<0.001) and higher perceived quality (B=0.220, p<0.001); female vs band showed higher perceived quality (B=0.128, p<0.001). Number of albums (log) negatively predicted engagement (B≈-0.039, p<0.001) and quality (B≈-0.059, p<0.001). Likes on comments (log) strongly increased engagement (B≈0.896, p<0.001) and quality (B≈0.284, p<0.001). - Song content: Lyric nostalgia positively predicted engagement (B≈0.027, p<0.01) and quality (B≈0.020, p<0.01). Lyric sadness was positively associated with both engagement (B≈0.005, p<0.001) and quality (B≈0.005, p<0.001). More negative lyric polarity slightly decreased engagement (B≈-0.001, p<0.01) but increased quality (B≈0.0004, p<0.01). - Comment topics (H1): Nostalgia in comments increased engagement (B=0.093, p<0.001) and quality (B=0.076, p<0.01). Comments about films/TV OSTs increased engagement (B=0.014, p<0.05) but reduced quality (B=-0.015, p<0.01). Expressing feelings about music reduced quality (B=-0.012, p<0.05). Evaluations of singer performance increased engagement (B=0.020, p<0.001). H1 supported. - Comment emotions (H2): Higher negative polarity in comments increased engagement (B=0.024, p<0.001) and quality (B=0.018, p<0.001). Love in comments increased engagement (B=0.036, p<0.001). For quality, sadness (B=0.029, p<0.10), anger (B=0.298, p<0.01), disgust (B=0.030, p<0.01), and surprise (B=0.148, p<0.05) were positive, while fear reduced quality (B=-0.065, p<0.05). H2 supported. - Matching/moderation (H3): Lyric nostalgia strengthened the effect of comment nostalgia on engagement (interaction B=0.030, p<0.05) and strengthened the effect of comment sadness on engagement (B=0.071, p<0.05). Negative comment polarity weakened the effect of comment nostalgia on engagement (B=-0.033, p<0.001) and on quality (B=-0.024, p<0.001). H3 partially supported. - Practical pattern: User comment content (autobiographical nostalgia and negative emotions) substantially amplifies song influence, especially aiding emerging artists with fewer albums/resources.

Findings support narrative transportation theory in a social media context: when music evokes autobiographical memories and emotions, users articulate these in comments, forming collective storytelling that raises engagement and perceived quality. The strong effects of nostalgic themes and negative emotions in comments suggest that emotionally resonant, memory-laden narratives drive attention and evaluation. Source-level inequalities persist (male and established artists fare better), yet user comments can partially offset disadvantages, benefiting less established artists by mobilizing active audience meaning-making. The results highlight that platformed cultural consumption is co-created by audiences whose comments function as narrators and cues, thereby extending the impact of content beyond intrinsic song features. Moderation analyses indicate that alignment between song themes (e.g., lyric nostalgia) and comment emotions/themes can further enhance engagement, whereas high negativity in comments can dampen the positive effect of nostalgic narratives. This advances theory by integrating audience co-creation and paratextual environments into narrative transportation, and it broadens empirical understanding in a non-Western platform setting.

The paper demonstrates that collective storytelling in user comments—especially autobiographical nostalgia and negative emotional framing—significantly elevates user engagement and perceived quality of songs on a major Chinese streaming platform. It advances narrative transportation theory by centering active audiences and paratextual comment spaces, and shows that comment dynamics can particularly boost the visibility of emerging artists. Future research should compare multiple streaming platforms, incorporate audio features (e.g., timbre, rhythm), and employ mixed methods (surveys, interviews) to capture broader user meaning-making. Cross-cultural analyses of foreign versus domestic songs could illuminate international flow and differences in engagement dynamics.

- Platform scope: Analysis is limited to NetEase Cloud Music; results may not generalize to other platforms with different affordances. - Omitted audio features: No direct analysis of musical/audio attributes (e.g., pitch, tone, rhythm) due to technical constraints. - Commenter bias: Focus on publicly posted comments may overrepresent users willing to share personal stories, not the full listener population. - Cultural scope: Restricted to Chinese songs; foreign music engagement and cross-cultural dynamics were not examined.