社群對話中的審議民主與政治極化：以新冠疫情記者會YouTube直播聊天室為例

Abstract

網路社群平台能否作為線上公共領域，讓人們不受限制地獲取資訊、不受制度影響平等地參與理性政治討論，是學界討論已久的議題：樂觀者認為，社群媒體重振了公共領域，使用者得獲得更多資訊近用權、公開發表意見的機會；悲觀者則認為，社群平台匿名性、去個體化特質，常促成攻擊性與仇恨言論，加上可能有受情緒所連結與動員的情感公眾、情緒渲染等現象，較難達到理性審議理想。其中，現今社群影音平台蔚為盛行，諸多使用者會於直播聊天室即時留言互動，然而相較其他平台，目前針對影音直播聊天室之輿論分析尚未成顯學，其參與者之行為模式與言論特徵值得進一步探索。

本研究運用文字探勘方法搭配內容分析，以新冠疫情記者會衛生福利部疾病管制署YouTube頻道直播聊天室為研究標的，旨在探索疫情記者會直播觀眾呈現的網路公眾與社群特性，以及直播聊天室作為線上審議空間之潛力與難處。此外，由留言中並觀察到相異政黨立場者爭執論戰之情形，並以政黨支持者代稱使用作為切入點，嘗試由政黨認同極化、負面黨性、群體衝突等概念詮釋解讀。

本研究發現，疫情記者會觀眾多為輕度參與、僅有少數熱衷者持續參與，曾於五場以上記者會留言者已是最積極參與的前10%用戶，亦有部分洗版者干擾討論秩序；留言內容以社交性、情緒性為多，雖有提出意見看法者但為少數；觀眾之間的回覆互動，直至疫情中後期始增加。至於直播聊天室中圍繞政黨立場的爭論，得發現自第一波本土疫情爆發後佔比上升，且在第二波本土疫情時期，執政黨受到的抨擊顯著增長。本研究之方法與技術上尚有未臻完美之處，仍可為後續從事YouTube直播聊天室分析、政治輿情分析研究者作為參考。

Whether online social media platforms can serve as a public sphere where individuals can access information without restrictions and participate in rational political discussions free from institutional regulation has long been debated. Optimists argue that social media revitalizes the public sphere, providing users with more information access and opportunities to express their opinions openly. Pessimists, on the other hand, believe that the anonymity and depersonalization of social media often lead to aggressive and hateful speech; besides, netizens often mobilized by emotions, makes it difficult to achieve the ideal of rational deliberation. Plenty of users engage in real-time interactions through live streaming chat rooms nowadays, however, compared to other social media platforms, the analysis of live chat rooms has not yet become a prominent field, and the behavioral patterns and speech features of participants are worthy of further exploration.

In this research, text mining methods combined with content analysis are used to explore the online public and community characteristics of the YouTube audience of the COVID-19 Press Conference held by Taiwan Centers for Disease Control (CDC). The potential and difficulties of the live chat room as an online deliberation space are investigated. Furthermore, the disputes between supporters of different political parties are evident in the live chat messages, thus the labels of political party supporters are used as a point of entry, and concepts such as political polarization, negative partisanship and group conflict are adopted to interpret and explain the phenomenon.

The research has found that most of the COVID-19 Press Conference audience in the live chat rooms are lightly involved, with only a few enthusiastic participants, those who have left messages at more than five press conferences are already the top 10% of the most active users, and there are also spammers who disrupt discussion order. The content of the chat messages is mostly social and emotional, few of them propose arguments. The conversations between the audience gradually increase since the later stages of the epidemic. As for the debates surrounding political stance, it is found that the proportion of disputes has increased since the first wave of the domestic epidemic; moreover, during the second wave of the domestic epidemic, there is a significant growth in criticisms of the ruling party. The research methods and techniques still have room for improvement, but they can serve as a reference for future researchers analyzing YouTube live chat rooms and political public opinion.