Abstract
Automated Emotion Recognition in text is a challenging application that has attracted significant interest. This paper addresses the significant need to study recently-introduced text corpora labeled for emotion recognition tasks. In this paper, we detail and analyze 30 text corpora introduced since 2018, offering insights into their sources, languages, emotion labels, annotation methodologies, availability, size, and other characteristics. We also summarize previous models and results on emotion recognition using the corpora in our study, and share recent events such as shared tasks and other resources such as lexicons. Finally, we offer a discussion of related practices and challenges. Our aim is for this paper and the accompanying online repository to function as a comprehensive resource, offering researchers a centralized hub for accessing related links, information, comparisons, and more, while also educating the research community on current practices for data collection, annotation, and usage in emotion recognition.