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

Abstract

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

Towards Sustainable Use of Machine Translation: Usability and Perceived Quality from the End-User Perspective

“…One of the wide areas of application of automated translation is user-generated content. Saadany et al [21] tested the ability of a machine translation online system to translate user-generated content. The researchers uncovered a variety of linguistic obstacles in translation and concluded that using neural machine translation technologies to translate raw texts could be harmful, as it could send users a message that is distinct from or even contradicts the intended meaning [21].…”

Discrimination on football Twitter: the role of humour in the Othering of minorities

“…The majority of tweets used in this study relate to the Premier League and men's football generally due to the continued growth of both racism and homophobia in the league as shown in Kick it Out's (2022) statistics. The football fandom is global, however, this study focuses on tweets in English due to potential linguistic issues with interpretation in the translation process (Saadany 2021). Discriminatory tweets were collected, typically from fans.…”

Challenges and Opportunities of Text-Based Emotion Detection: A Survey

“…The manual analysis has revealed that many idioms were translated literally, which not only affected the source text's ability to retain sentiment but also frequently resulted in the nonsensical target text. The same problem exists in other code-mixed or low-resources datasets [258]. • Cross-domain emotion detection: The basic tenet of the cross-domain emotion analysis method is that given enough words that convey a range of emotions from multiple domains, emotions contained in current comment information can be swiftly and accurately identified.…”