Challenges in Translation of Emotions in Multilingual User-Generated Content: Twitter as a Case Study

Although emotions are universal concepts, transferring the different shades\nof emotion from one language to another may not always be straightforward for\nhuman translators, let alone for machine translation systems. Moreover, the\ncognitive states are established by verbal explanations of experience which is\nshaped by both the verbal and cultural contexts. There are a number of verbal\ncontexts where expression of emotions constitutes the pivotal component of the\nmessage. This is particularly true for User-Generated Content (UGC) which can\nbe in the form of a review of a product or a service, a tweet, or a social\nmedia post. Recently, it has become common practice for multilingual websites\nsuch as Twitter to provide an automatic translation of UGC to reach out to\ntheir linguistically diverse users. In such scenarios, the process of\ntranslating the user's emotion is entirely automatic with no human\nintervention, neither for post-editing nor for accuracy checking. In this\nresearch, we assess whether automatic translation tools can be a successful\nreal-life utility in transferring emotion in user-generated multilingual data\nsuch as tweets. We show that there are linguistic phenomena specific of Twitter\ndata that pose a challenge in translation of emotions in different languages.\nWe summarise these challenges in a list of linguistic features and show how\nfrequent these features are in different language pairs. We also assess the\ncapacity of commonly used methods for evaluating the performance of an MT\nsystem with respect to the preservation of emotion in the source text.\n