서사를 넘어: 유튜브 광고에서의 감정의 시간적 역동성과 인지적 주제 = Beyond the Narrative: Temporal Emotion Dynamics and Cognitive Themes in YouTube Advertising

다국어 초록 (Multilingual Abstract)

In highly participatory digital video platforms, advertising effects are no longer limited to measurements at the level of attitudes or memory, but are increasingly reflected in emotional reactions, discussion content, and potential behavioral expressions observed in comment sections. On platforms such as YouTube, the large volume of comments generated after advertisement releases provides an opportunity to observe audience responses in real and natural contexts. Although narrative advertising is widely regarded as having stronger emotional appeal, whether it necessarily performs better than non-narrative advertising in real social media environments remains to be further examined. Based on advertising comment data from YouTube, this study compares narrative and non-narrative advertisements in terms of emotional responses, discussion topics, and behavioral intention expressions, and examines how these responses change across different post-release stages. Forty brand advertisement videos were selected for analysis, including 20 narrative and 20 non-narrative advertisements, with advertisement types identified through a double-coding procedure. After data cleaning and preprocessing, a total of 151,051 valid comments were obtained. While preserving the chronological order of comments, the data were divided into early, middle, and late stages to examine temporal variations in audience responses. Methodologically, this study applies the VADER sentiment analysis model to measure emotional polarity in comments and uses Latent Dirichlet Allocation (LDA) to identify major discussion topics in comments on different advertisement types. In addition, purchase intention and recommendation intention are identified through a keyword-matching approach, and the relationships between emotional categories and behavioral intentions are further analyzed. The results show that: (1) narrative advertisements consistently display higher levels of positive emotional expression, whereas non-narrative advertisements exhibit relatively higher proportions of negative emotion; (2) for both advertisement types, emotional intensity as well as the proportions of positive and negative emotions gradually decrease over diffusion stages; (3) comments on narrative advertisements mainly focus on story contexts, character relationships, and emotional experiences, whereas comments on non-narrative advertisements concentrate more on product attributes, performance, and price–value evaluations, and the topic structures of both advertisement types remain generally stable over time; (4) non-narrative advertisements show a higher proportion of purchase intention expressions, while narrative advertisements show a relative advantage in recommendation intention; and (5) recommendation intention varies more noticeably across emotional categories, whereas purchase intention shows a relatively stable distribution across different emotional contexts. By analyzing social media comment data, this study identifies differences in audience responses to narrative and non-narrative advertisements on digital platforms, and provides empirical observations based on naturalistic data.

목차 (Table of Contents)

• 1. Introduction 1
• 2. Theoretical Background 5
• 2.1. Definition and Core Elements of Narrative Advertising 5
• 2.2. Mechanisms of Narrative Advertising 6
• 2.3. Definition of Non-Narrative Advertising 7
• 2.4. Cognitive Processing Mechanisms of Non-Narrative Advertising 8
• 2.5. UGC Comments and Online Responses 9
• 2.6. Behavioral Intention Expressions in Social Media Comments 10
• 3. Hypothesis Development 12
• 4. Research Methodology 17
• 4.1. Research Framework 17
• 4.2. Data Collection 18
• 4.3. Data Preprocessing and Time Segmentation 20
• 4.4 Sentiment Analysis 21
• 4.5. Topic Modeling 22
• 4.6. Temporal Analysis of Topic Distributions 23
• 4.7. Behavioral Intention Analysis 24
• 5. Analysis Result 26
• 5.1. Sentiment Analysis 26
• 5.2. Topic Modeling (LDA) 30
• 5.3. Behavioral Intention analysis 36
• 5.4 Association Between Emotional Types and Behavioral Intentions 42
• 6. Conclusion 45
• 6.1. Discussion 45
• 6.2. Implications 46
• 6.3. Limitations and Future Research 48
• Reference 50