This project includes a working paper developing and testing a new discrete emotions dictionary (DED), "Constructing a Dictionary for the Automated Identification of Discrete Emotions in News Content," by Sarah Fioroni, Ariel Hasell, Stuart Soroka and Brian Weeks.
The dictionary is developed and tested to be used for news content. It provide a conservative estimate of the appearance of four discrete emotions in text. It can be used either on its own, or as a first step towards finding a sufficiently 'balanced' sample for use in machine learning.
The dictionary itself is stored in four plain text files, also available here.