Abstract
The emotional connotation attached to words undergoes language change. Inthis paper, we aim at estimating the emotion which is associated with a givenword in former language stages of English and German. Emotion is representedfollowing the popular Valence-Arousal-Dominance (VAD) annotation scheme. Whilebeing more expressive than polarity alone, existing word polarity inductionmethods are typically not suited for addressing it. To overcome thislimitation, we present adaptations of two popular algorithms to VAD. To measurethe effectiveness of these methods in diachronic settings, we present two smallgold lexicons for historical word emotions which where annotated by scholarswith proficiency in the respective language (stage). In contrast to claims inprevious work, we find that limiting the training set to words of supposedlystable emotional meaning does in fact decrease performance, rather thanboosting it. This contribution thus elaborates algorithmic and experimentalresults which form the basis of some of our later publications (Buechel et al.,2017; Buechel and Hahn, 2018; Hellrich et al., 2018) yet had to be skippedthere, partly for brevity, partly for a change in focus of our work.