The classification of emotions from text is usually done on basis of a single psychological model of emotion and affect with a singletask classifier. The goal of this bachelor thesis is to evaluate the merit of multitask learning across emotion models. Here, a multitask approach is taken, where multiple data sets, each annotated on basis of a different psychological model, are trained in a multiheaded neural network. Comparing the F1-score / mean-squared-error of the multitask classifiers to the singletask classifiers show no improvement.