Summary: Researchers have advanced our understanding of universal emotions across languages by applying colexification analysis, a linguistic method that traces when a single word expresses multiple related concepts. Their cross-linguistic network study highlights four emotion-related concepts—“GOOD,” “WANT,” “BAD,” and “LOVE”—as central hubs that show the most associations with other emotion words across diverse languages.
This result complements findings from established semantic approaches, including the natural semantic metalanguage (NSM), reinforcing the idea that certain emotional concepts are widely shared across languages and cultures. The study’s insights have practical relevance for natural language processing (NLP), sentiment analysis, and the ongoing development of large language models (LLMs), helping these systems better recognize and handle emotion-related meaning in text.
Key Facts:
- The colexification analysis identified “GOOD,” “WANT,” “BAD,” and “LOVE” as central emotion concepts that are highly connected to other emotion terms across languages.
- Three of these four core concepts correspond to semantic primes identified by traditional semantic methods and NSM, supporting their cross-linguistic universality.
- Findings have potential applications in natural language processing, sentiment analysis, and the design of language models aimed at improving cross-cultural online communication.
Source: Tokyo University of Science
Emotions shape thought and behavior, so identifying which emotional concepts serve as central hubs in meaning is valuable both scientifically and practically. Businesses, non-profits, and communicators can benefit from knowing which emotional concepts are most salient across languages when crafting messages intended to resonate across cultural boundaries.
Colexification occurs when a single lexical form covers multiple semantically related concepts—an indirect window into how languages organize meaning. For example, one word in a language might convey both “BAD” and “SEVERE,” indicating those concepts are closely linked in that language’s semantic system. By mapping these overlaps across many languages, colexification analysis constructs networks that reveal which concepts function as semantic hubs.
In a study published online in Scientific Reports on December 09, 2023, researchers from Japan used a colexification network approach to examine emotion-related vocabulary across languages. The author team—led by Dr. Tohru Ikeguchi, with Ms. Mitsuki Fukuya, Dr. Tomoko Matsumoto (Tokyo University of Science), and Dr. Yutaka Shimada (Saitama University)—built networks in which nodes represent concepts and edges represent colexification links. Edge weights captured the strength of colexification between concept pairs, enabling the identification of highly connected “hub” emotions.
“Colexification is when a single word covers multiple concepts. For example, the Spanish word ‘malo’ can mean both ‘BAD’ and ‘SEVERE,’” explains Dr. Ikeguchi. “By focusing on colexification, we could detect central emotions that are semantically close to many other emotion concepts across languages.”
The analysis singled out four central emotion concepts—“GOOD,” “WANT,” “BAD,” and “LOVE”—as having the largest combined weights and hub-like connectivity in the colexification network. Notably, three of these—“GOOD,” “BAD,” and “WANT”—align with semantic primes identified by NSM researchers who used traditional semantic methods to study many languages. That convergence supports the interpretation that these core concepts are not limited to a single language but reflect broader, cross-linguistic semantic structure.
These findings shed light on how emotional meaning may be organized in human language, with implications for the study of language evolution and cross-cultural communication. Because words mediate emotional expression and understanding, mapping their shared semantic links can reveal enduring patterns in human cognition and communication.
From a technological perspective, improved knowledge of emotion hubs in language benefits natural language processing and sentiment analysis. Identifying concepts that frequently colexify with positive or negative affect can refine sentiment classifiers, improve emotion recognition, and help LLMs produce or interpret emotionally nuanced text more reliably. As organizations and researchers invest in language technologies, insights from cross-linguistic semantic networks can guide the development of more culturally aware NLP systems.
About this language and emotion research news
Author: Hiroshi Matsuda
Source: Tokyo University of Science
Contact: Hiroshi Matsuda – Tokyo University of Science
Image: The image is credited to Neuroscience News
Original Research: Open access.
“Central emotions and hubs in a colexification network” by Tohru Ikeguchi et al. Scientific Reports
Abstract
Central emotions and hubs in a colexification network
By focusing on colexification, we detected central emotions that share semantic commonalities with many other emotions through relationships of similarity and association. We constructed colexification networks from multiple languages by representing concepts as vertices and colexification links as edges. We then identified emotion concepts with high aggregated weights and located hub emotions within these networks. The analysis produced four central emotion concepts: “GOOD,” “WANT,” “BAD,” and “LOVE.”