新媒體與傳統媒體對於閱聽人行為的影響：以YouTube上「博愛座」報導為例

Abstract

在數位媒體快速發展的當代社會，傳統媒體與新媒體在議題再現與閱聽人互動上展現出顯著差異。本研究以2024年6月博愛座衝突事件為背景，探討不同媒體類型如何透過差異化的框架策略影響閱聽人的認知與行為反應。研究採用媒體框架理論為基礎，結合計算傳播學方法，對YouTube平台上與「博愛座」議題相關的影片與留言進行系統性分析。隨著數位媒體生態重塑，傳統新聞媒體面臨前所未有的轉型挑戰。為因應閱聽人收視習慣改變與平台競爭壓力，傳統新聞媒體積極佈局YouTube平台，建立官方頻道延伸影響力。本研究觀察發現，傳統新聞媒體如TVBS、民視等，在YouTube平台上維持原有新聞專業標準，以記者採訪與主播播報為主要敘事模式，保持客觀報導立場與完整剪輯結構。然而，在數位原生內容競爭下，傳統媒體必須在維持新聞品質與迎合平台演算法間取得平衡，此一轉型過程反映了媒體產業在數位時代的適應性變革，也為本研究提供豐富的媒體類型比較基礎。本研究收集2018年7月1日至2024年7月1日期間YouTube平台上634部博愛座相關影片與37,997則留言，運用自然語言處理技術、LDA主題建模、BERTopic分析與GPT-4情感分類等方法進行文本分析。研究將影片分類為六種媒體類型：傳統新聞媒體、KOL頻道、直播影片、數位原生媒體、名嘴談話節目與街頭訪問，並識別出三種主要框架：世代衝突框架、同理心訴求框架與資源配置框架。創新性地引入時序分析方法，建立半數集中時間點、九成集中時間點與互動峰值時間點等測量指標，量化不同媒體類型的傳播軌跡特徵。研究發現，傳統新聞媒體雖在影片數量上佔主導地位（63.72%），但在互動密度方面不及新媒體形式。街頭訪問與直播影片平均每部引發223.65與162.07筆留言，遠高於傳統新聞的46.45筆。情感分析顯示，新媒體內容更易引發負面情緒反應，街訪與直播影片的負面情緒比例分別達49.6%與53.7%。時序分析揭示三種傳播模式：即時爆發型（名嘴、街訪、KOL影片在第1小時內達峰值）、快速反應型（傳統新聞與直播在第1小時達峰值）、延遲擴散型（數位媒體在第16小時達峰值）。主題分析發現，不同媒體類型在框架運用上存在系統性差異：新媒體偏好採用世代衝突框架，傳統媒體傾向使用同理心與資源配置框架。本研究的理論貢獻在於豐富了媒體框架理論在數位環境中的解釋力，將時間維度納入框架效果分析，深化對媒體影響力動態演變過程的理解。方法論上，建立了適用於中文社群媒體的大數據分析流程，創新性地將時序分析納入框架效果研究。實務意涵方面，研究結果顯示不同媒體類型具有差異化的影響模式，為媒體從業者與政策制定者提供重要參考，建議在處理敏感公共議題時應審慎運用框架策略，避免過度情緒化的敘事手法，促進理性公共討論。

In contemporary society with rapid digital media development, traditional media and new media demonstrate significant differences in issue representation and audience interaction. This study examines how different media types influence audience cognition and behavioral responses through differentiated framing strategies, using the priority seat conflict incident in June 2024 as background. Based on media framing theory and incorporating computational communication methods, this research conducts systematic analysis of videos and comments related to the"priority seat" issue on the YouTube platform. This study collected 634 priority seat-related videos and 37,997 comments from YouTube spanning July 1, 2018, to July 1, 2024, employing natural language processing techniques, LDA topic modeling, BERTopic analysis, and GPT-4 sentiment classification for textual analysis. Videos were categorized into six media types: traditional news media, KOL channels, live streaming videos, digital native media, talk shows, and street interviews. Three primary frames were identified: generational conflict frame, empathy frame, and resource allocation frame.Innovatively introducing temporal analysis methods, the study established measurement indicators including 50% concentration time point, 90% concentration time point, and peak interaction time point to quantify the transmission trajectory characteristics of different media types.The findings reveal that while traditional news media dominates in video quantity (63.72%), it falls behind new media forms in interaction density. Street interviews and live streaming videos generated an average of 223.65 and 162.07 comments per video respectively, significantly higher than traditional news media's 46.45 comments. Sentiment analysis shows new media content more readily triggers negative emotional responses, with street interviews and live streaming showing negative emotion ratios of 49.6% and 53.7% respectively. Temporal analysis reveals three transmission patterns: immediate burst type (talk shows, street interviews, KOL videos peak at hour 0), rapid response type (traditional news and live streaming peak at hour 1), and delayed diffusion type (digital media peak at hour 16). Topic analysis found systematic differences in frame utilization across media types: new media prefer generational conflict frames, while traditional media tend toward empathy and resource allocation frames.The theoretical contribution of this study lies in enriching the explanatory power of media framing theory in digital environments and incorporating temporal dimensions into framing effect analysis, deepening understanding of the dynamic evolution process of media influence. Methodologically, it establishes a big data analysis framework suitable for Chinese social media and innovatively incorporates temporal analysis into framing effect research. Regarding practical implications, the results demonstrate that different media types possess differentiated influence patterns, providing important references for media practitioners and policymakers. The study recommends cautious use of framing strategies when handling sensitive public issues, avoiding overly emotional narrative approaches to promote rational public discourse.