蔡依霖

Yi-Lin Tsai

集體用戶表現、意見氣候、準統計官能、語料庫分析

aggregate user representation, opinion climate, quasi-statistical organ, corpus analysis

本文重新解構Noelle-Neumann在沉默螺旋理論中的要素「準統計官能」（Quasi-Statistical Organ），將原本理論所處的媒介環境修正為當今多元的資訊環境，透過平臺的特性與介面展現的集體用戶表現（Aggregate User Representation）發展出更貼近批踢踢實業坊（PTT）使用者「鄉民」視角的資料蒐集途徑，並利用語料庫分析取徑萃取出能夠提取、預測意見氣候的探測語（Probes），藉此發展出能夠不先預設觀測範圍的意見氣候分析架構。

資料結果顯示，本文還原準統計官能的原理，利用介面的集體用戶表現與修正過後的「探測語」概念設計出可以持續更新的資料蒐集方式，並搭配自建語料庫分析的方式，能建立一個可持續更新探測語的分析架構，且實測以此架構找出的探測語確實能發揮預測意見氣候的效果。本文認為從介面與準統計官能的角度建立的意見氣候分析框架，將更能夠因應瞬息萬變的網路資訊環境，且探索過去預先劃定觀測範圍所不能觀測到的細微現象。

This paper revisits Noelle-Neumann’s Spiral of Silence theory, proposed in 1972, and identifies areas that need enhancement in the context of the contemporary information environment. By utilizing the Quasi-Statistical Organ from the theory, the study establishes a more suitable analytical framework for modern public opinion research. The following questions are explored. (1) How does the Quasi-Statistical Organ operate within the new media environment? (2) How can data be collected according to the logic of the Quasi-Statistical Organ? (3) Can observation methods designed based on the Quasi-Statistical Organ predict emerging popular topics?

The literature is divided into three sections: Theoretical Gaps Due to Changes in Media Environment, Social Cues in the New Media Environment, and New Research Approaches Combining New Media Opinion Climate and Corpus Methods.

First, the strand of Theoretical Gaps Due to Changes in Media Environment explores domestic and international studies on the application of the Spiral of Silence theory. It finds that changes in the information environment have enabled audiences to exhibit more diverse responses. Whereas in the past, audiences could only speculate in interviews whether they would express their political choices to strangers on a train or participate in opinion polls, today, they can obtain social cues from various sources. Notably, the process of browsing social media allows audiences to piece together aggregate user representations (AURs), enabling them to imagine the majority opinion and decide whether to express their own views or subtly blend into AUR (Walther & Jang, 2012). Audiences now have more agency in the formation of opinion climate. More importantly, the behavior of detecting opinion climates through AUR helps researchers clarify the operation of the Quasi-Statistical Organ, a principle previously unexplained in the Spiral of Silence theory, in the contemporary media environment.

Second, the section Social Cues in the New Media Environment examines how, in different information environments, do proprietor content (PC), aggregate user representations (AURs), and user-generated content (UGC) constitute social cues from the audience’s perspective.

Lastly, the part New Research Approaches Combining New Media Opinion Climate and Corpus Methods reviews research on public opinion to determine if there are methodological limitations, such as initiating studies from individual cases or topics, that could miss unanticipated research outcomes due to top-down research designs. This study argues that incorporating corpus methods, which include audience behavior of collecting opinion climates using AUR, into the data collection process and establishing a continuously updating research framework through linguistic probes align researchers’ perspectives more closely with those of the audience.

This study selects the Gossip Board of PTT.cc, Taiwan’s most popular bulletin board system, as the data source, using the aggregate user representations (AUR) accessible to users as the data collection standard. It collects data from January to June 2022 on PTT’s Gossip Board, specifically targeting posts that (1) have a net positive score (upvotes minus downvotes) greater than 100 and (2) have titles containing “Re: [新聞]” (Re: [News]), including any associated comments. The collected data comprise 651 posts and 319,876 comments. The data are organized and analyzed using the Chinese corpus analysis tool CORPRO. The criterion of selecting posts with a net positive score greater than 100 ensures that these posts are marked as “爆!” (explosive), making them more visible to users. Additionally, the requirement that the title contains “Re: [新聞]” indicates that the content is reposted news that has been brought back for discussion. This behavior is akin to audiences simultaneously looking towards the public domain and the community, receiving public news information, and then sharing it with others who should see it. This thought process closely aligns with the intuitive way audiences gauge others’ opinions on issues, as described in the Quasi-Statistical Organ. Moreover, the study builds upon Johansen’s (2017) method of using corpora to identify specific function words in the form1-to-function-to-form2 approach in probe studies, allowing researchers to continuously update probes when identifying emerging opinion climates.

The results of this paper establish a sustainable and topic-independent opinion analysis framework. First, the word frequency of a large number of comments is calculated. Second, replies to less popular posts are collected and subjected to keyness calculation to identify which words appear exclusively in popular posts. This set of words serves as probes capable of extracting the opinion climate. Third, articles containing these probes are searched over the following week, revealing the opinion climate currently under heated discussion. The research findings note that the analytical framework built using probes is indeed effective when applied to the six months of data collected herein.

The purpose of this study is to address gaps in the description of the Quasi-Statistical Organ principle within the Spiral of Silence theory and to establish an opinion climate analysis framework that transcends time and topics, thus aligning closer to the perspectives of online community members. The primary research contributions include outlining the operational principles of the Quasi-Statistical Organ in the contemporary media environment and proposing a novel analytical framework for opinion climate analysis. This paper first addresses the deficiencies in the Spiral of Silence theory by outlining the operational principles of the Quasi-Statistical Organ in the current media environment and then presents a new, sustainable framework for opinion climate analysis that extends beyond predetermined research boundaries.

Based on this paper, future research can explore comparative studies on other platforms or serve as a reference for information literacy research. The limitations of this study include the inability to cover the impact of other social media interfaces and the inability to exclude any influences of some extreme cases on the overall data.