Velina Slavova, Filip Andonov
New Bulgarian University
https://doi.org/10.53656/math2024-2-6-inv
Abstract. This paper is based on the results of published authorial research on the relationship between the sound composition of language and emotion, leading to the conclusions and generalizations drawn here.
In order to uncover deeper relationships between language and emotion, a methodology and corresponding automated system was developed that allowed the application of statistical methods over the phonemic composition of large volumes of textual data. So far, the English data and the results of the applied statistical methods confirm the existence of dependencies between phoneme composition expressed as vowel-consonant pairs and emotional valence (positive-negative emotion) for news texts
from the Internet.
The presented research necessitated the development of an automated system. The system was implemented with the participation of computer science students, within the framework of full-time courses. Our experience suggests that incorporating research topics into the undergraduate curriculum helps to understand both the role of in formation technology for research and the limitations of systems used as “artificial intelligence”.
Keywords: phonosemantics; emotional sound symbolism; emotion recognition in text; education in informatics
