Twitter Mentions of February 6th Earthquake: An Emotional Language Analysis
Cemile UzunSocial media is crucial during and after earthquakes, as it enables people to both receive information and express themselves. This study focuses on the analysis of the emotional language used by people when sharing their experiences related to the earthquake that occurred in Turkey on February 6, 2023. In the current study, the emotional language analysis focused on the social media tool Twitter alone. Emotion analysis is a natural language processing technique used to identify emotions expressed in a text. The current study aims to reveal the reflection of emotions about the earthquake in specific expressions and the variation in these emotions over time. For this purpose, an emotional language analysis of tweets posted on Twitter with the hashtag #earthquake between February 6 and June 6 was conducted. Nakamura’s system of emotion analysis was used to perform the emotional language analysis. In his work “Kanjō hyōgen jiten (Dictionary of emotive expressions),” Nakamura emphasized ten classifications of emotions: excitement, embarrassment, joy, fondness, dislike, sadness, anger, surprise, fear, and relief. From these, six emotions, namely joy, grief, anger, surprise, fear, and relief, were selected to analyze the emotional language in tweets about the earthquake. Our study is distinct from other studies in Turkey in terms of both revealing the emotional language resulting from the earthquake and using Nakamura’s emotion analysis.
6 Şubat Depremi ve Twitter Duygu Analizi
Cemile UzunSosyal medya, insanların hem bilgi almasını hem de kendilerini ifade etmesini sağladığı için deprem anında ve sonrasında önemli bir yere sahiptir. Çalışmada, insanların depreme ilişkin tecrübelerini ifade ederken kullandıkları duygu analizi dili üzerinde durulmuştur. Depremle ilgili duygu analizi dili tespit edilirken sosyal medya aracı olarak sadece Twitter kullanılmıştır. Duygu analizi, bir metinde ifade edilen duyguyu belirlemek için kullanılan doğal dil işleme tekniğidir. Çalışmanın amacı depreme yönelik duyguların dile nasıl yansıdığını ve bu duyguların zamanla nasıl bir değişim izlediğini ortaya çıkarmaktır. Bu amaçla, Twitter’de deprem etiketiyle 06 Şubat-06 Haziran tarihleri arasında atılan tivitlerdeki duygu dili analizi incelenmiştir. Duygu analizi dilini tespit etmek için Nakamura’nın duygu analizi sistemi kullanılmıştır. Nakamura, “Emotional Display Dictionary” adlı çalışmasında “heyecan, utanç, sevinç, düşkünlük, hoşlanmama, üzüntü, öfke, sürpriz, korku ve rahatlama” olmak üzere on duygu analizi sınıflandırması üzerinde durmuştur. Depremin insanlar üzerinde yarattığı duygular göz önünde bulundurularak Nakamura’nın duygu analizi depreme göre uyarlanmış ve 06 Şubat duygu dili analizi tespit edilirken “sevinç, keder, öfke, şaşırma, korku, rahatlama” olmak üzere altı duygu dili analizi kullanılmıştır. Hem Türkiye’de yaşanmış bir depremin duygu dili analizini ortaya koyması hem de Nakamura’nın duygu analizini kullanması yönüyle çalışmamız Türkiye’deki diğer çalışmalardan ayrılmaktadır.
An earthquake is a sudden release of energy in the rocks of the Earth that generates seismic waves. Turkey is a seismically active country that has been subjected to severe earthquakes throughout history. Turkey has experienced earthquakes of magnitude 7 and above since the 1500s. On February 6, 2023, two separate earthquakes of magnitudes 7.7 and 7.6 occurred in the Pazarcık district of Kahramanmaraş. This event was one of the earthquakes that deeply affected the history of Turkey.
Social media are online platforms where users can create content, share it, and interact with each other. During a disaster, a substantial amount of data were generated by social media users. When an earthquake occurs, social media acts as a crucial source of information for decision-making support systems. Furthermore, the experiences of people living through an earthquake can be quite diverse. There are several social media tools through which people share their experiences. Emotion analysis, a technique used in the field of natural language processing, is an analytical approach to identifying emotional responses in a data type. In this study, the flow of information on Twitter, including the moment of the earthquake and its aftermath, was investigated using emotional language analysis to assess the emotional language used by people in the tweets.
The data analyzed in this study were limited to tweets posted between February 6 and June 6, 2023. Only those tweets talking about the primary earthquake and the major aftershocks were considered, and the analyzed tweets were further limited to the keywords #earthquake and #zelzele. The data collection method used in this study was based on A. Nakamura’s (2005) emotion analysis. These emotions were adapted to those that can be felt during an earthquake. The major classes of emotions selected for the analysis were joy, shame, grief, anger, surprise, fear, and relief. The chief conclusions were as follows:
I. The most frequently used emotional language since the first moment of an earthquake was grief.
II. Emotions of fear and anger were more common in tweets posted in the afternoon. Furthermore, the emotional language of grief and fear seen in the early hours and days of the earthquake turned into the emotional language of relief over time.
III. The most commonly used emotional language in tweets after that of grief was the emotional language of relief. The emotion of relaxation was more common in tweets sent during the evening.
IV. Since March, there have been tweets expressing feelings of joy. In sentences using the language of joy, expressions describing the happiness of children have been encountered.
V. The least used emotional languages were joy, surprise, and shame.
It was observed that certain distinct emotions emerged in the sentences, despite the absence of explicit emotional language. The frequency of certain words, excluding those used in the emotional language analysis, was also determined. It was determined that “begging” was the most frequently used emotion, other than the ones included in the emotion analysis language, and the most frequently used word was the verb “to protect.”
