Latest News


Conceptual analysis of multiple languages has identified four central human emotions—GOOD, WANT, BAD, and LOVE


By analyzing emotional concepts in multiple languages using a colexification network, a research group made up of Professor Tohru Ikeguchi of the Department of Information and Computer Technology at the Faculty of Engineering, Graduate Student Mitsuki Fukuya of the Graduate School of Engineering and Associate Professor Tomoko Matsumoto of the Institute of Arts and Sciences at Tokyo University of Science, and Associate Professor Yutaka Shimada of the Department of Information and Computer Sciences, Graduate School of Science and Engineering at Saitama University, has identified four central emotions closely related to human emotional concepts: "GOOD," "WANT," "BAD," and "LOVE." The article was published in Scientific Reports dated December 9, 2023.

Emotions and emotion-related concepts are considered to play an important role not only in human communication but also in the field of natural language processing, in which text data are processed by computers using normal human language, particularly in sentiment analysis. The sentiment analysis methods, by which emotions and opinions are extracted and analyzed from sentences or speeches, enable semantic identification of sentiments, such as "positive," "negative," or "neutral" opinions, from text and sound information. These methods are useful in, for example, SNS monitoring and product review analysis.

The research group has focused to date on the structural analysis of languages from a network theory perspective. In this study, the research group attempted to extract central emotions that have semantic similarity with various other emotions by focusing on the colexification of central emotions for all languages combined. Using a colexification dataset, they examined whether capturing the colexification network structure could enable for clarification of central emotions common for all human beings.

Colexification is the phenomenon whereby a single word is associated with multiple concepts; for example, the Spanish word "malo" can have two meanings: "BAD" and "SEVERE." This indicates how the target language captures and expresses concepts.

When textual analysis of a language is performed, a large corpus that enables language processing (a database that includes a large collection of numerous examples of natural language and is tailored to be in the form for computer processing) needs to be collected. In contrast, an analysis using colexification does not require the collection of similarity data between languages because it uses existing translation dictionaries and the like. Therefore, colexification analysis seems to be a newer linguistic method that can analyze indirect semantic similarity.

The following network analysis was performed using an online database (CLICS3) on colexification of more than 3,000 languages. Colexification networks were created from multiple languages by first assigning a concept of emotion to a vertex and then setting up an edge based on the presence or absence of colexification between each vertex. A random walk approach was applied to the colexification network to evaluate the similarity between concepts.

As a result, four concepts with a large weight of edge, that is, the four central concepts, were identified. They are very relevant to other emotional concepts and can be interpreted as hubs in the colexification network. It was also found that many emotional concepts relevant to the layers close to the hubs belong to the same community as the hubs. The developed approach could offer a non-conventional perspective to facilitate the identification of concepts that serve as bridges between emotional concepts.

Ikeguchi commented, "In this study, we applied network theory to analyze the information from the colexification datasets, and by targeting concepts related to emotions, we found that the four central emotions were GOOD, WANT, BAD, and LOVE." The colexification datasets contain a variety of non-emotional concepts in different languages and language families. By applying network analysis techniques to these various concepts, as well as different languages and language families, I believe that we will be able to clarify how languages have developed, and how historical and geographical factors have affected this process."

Journal Information
Publication: Scientific Reports
Title: Central emotions and hubs in a colexification network
DOI: 10.1038/s41598-023-48922-8

This article has been translated by JST with permission from The Science News Ltd. ( Unauthorized reproduction of the article and photographs is prohibited.

Back to Latest News

Latest News

Recent Updates

    Most Viewed