Measuring Cultural Relativity of Emotional Valence and Arousal using Semantic Clustering and Twitter

Eugene Yuta Bann,Joanna J. Bryson

DOI: https://doi.org/10.48550/arXiv.1304.7507

2013-04-29

Abstract:Researchers since at least Darwin have debated whether and to what extent emotions are universal or culture-dependent. However, previous studies have primarily focused on facial expressions and on a limited set of emotions. Given that emotions have a substantial impact on human lives, evidence for cultural emotional relativity might be derived by applying distributional semantics techniques to a text corpus of self-reported behaviour. Here, we explore this idea by measuring the valence and arousal of the twelve most popular emotion keywords expressed on the micro-blogging site Twitter. We do this in three geographical regions: Europe, Asia and North America. We demonstrate that in our sample, the valence and arousal levels of the same emotion keywords differ significantly with respect to these geographical regions --- Europeans are, or at least present themselves as more positive and aroused, North Americans are more negative and Asians appear to be more positive but less aroused when compared to global valence and arousal levels of the same emotion keywords. Our work is the first in kind to programatically map large text corpora to a dimensional model of affect.

Computation and Language,Artificial Intelligence

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to explore the universality and cultural relativity of emotions. Specifically, researchers hope to measure whether there are significant cultural differences in the valence and arousal of these emotional keywords by analyzing the usage of emotional keywords on Twitter in different geographical regions (Europe, Asia, and North America). This research aims to provide new evidence for the cultural relativity of emotions by applying distributional semantic techniques to analyze large - scale text data. The paper points out that although previous research has mainly focused on facial expressions and a limited set of emotions, a more comprehensive understanding of cultural differences in emotions can be achieved by analyzing self - reported behaviors on social media.

The main contribution of the research lies in the first systematic use of "online" and "big data" methods to map large - scale text corpora into the core - dimensional model of emotions, thereby quantifying the differences in the valence and arousal of the same emotional keywords in different cultural backgrounds. This not only helps to understand the cultural dependence of emotional expression, but also provides new perspectives and methodological support for cross - cultural emotion analysis.