Mapping neologism and collective human behavioral changes : a study of COVID-19 related emergent neologisms using big data

Abstract

Neologisms have been widely recognized as an extremely sensitive linguistic indicator of a new social event. Prior studies have found that the preference for selecting a gain or loss framing strategy to represent a new social event such as a pandemic can respond to relevant policies, pandemic status, and speakers' emotional change during the pandemic. However, existing studies on disease-related neologisms largely ignored the important role of neologisms to respond to human behavioral changes during the epidemic. As an emerging pandemic, COVID-19 is not only a medical event but also a significant social event that generates a large number of online discussions. Under the big data era, a diachronic tracking study on the use of COVID-19 emergent neologisms on the internet can provide a critical lens for us to better understand collective human behavior during different stages of the pandemic. In this study, collective human behavior is operationalized as relevant policy announcement, pandemic development, and public emotional changes under the ongoing disease. This thesis employs a mixed research design. The qualitative part tracks the developmental patterns of COVID-19 emergent neologisms from the Baidu Index, from the end of 2019 to the end of March 2021, and analyzes that their fluctuations according to important policies. The most important part of the present thesis lies at the prediction and N-gram co-occurrence. The prediction involves two steps: (i) correlating the internet searches of COVID-19 emergent neologisms with the pandemic cases, and then (ii) training/validating/testing their mathematical relationship based on multiple (non)linear and fine-tuned regression models. To highlight a more important role of emergent neologisms in associating with the collective human behaviour, the current thesis compares the predictability of COVID-19 emergent neologisms with buzzwords motivated by the COVID-19 pandemic (i.e., vector names and Personal Protection Equipment (PPE) names). The other quantitative part explores if emergent neologisms can be good indicators to monitor the change of public attention. This part also involves two steps: (i) hypothesizing the public emotional change at different stages of the pandemic from the general development of COVID-19 emergent neologisms, and then (ii) verifying the hypothesis by N-gram co-occurrence (N = 2 - 6 Chinese words) of the crawled Sina Microblog posts. The qualitative result showed that the development of COVID-19 emergent neologisms corresponded with the important policies over the fifteen months after the pandemic outbreak. The prediction result showed that, compared with buzzwords, COVID-19 emergent neologisms are better to predict pandemic cases by using binomial main effects. A better fitting curve is a Least Angle Regression model. For the monitoring effect of COVID-19 emergent neologisms on the public emotional change, the public emotion was hypothesized and verified to experience the three stages from fear to relaxation together with caution based on observing the general development of COVID-19 emergent neologisms over the fifteen months. The contributions of this thesis are twofold. For real life application, emergent neologisms, rather than buzzwords, are better indicators for predicting an emerging public health event and for monitoring the change of public emotion. For theoretical explication, the current thesis proposes an interactive network associating emergent neologisms, relevant policies, pandemic development, and public attention to highlight the important role of emergent neologisms in responding to the collective human behavior and to provide valuable insights to the role of neologisms in language change.