Scientists have found that different languages develop according to similar rules
The scientists analysed 22 languages and concluded that dictionaries do not develop by chance. New words and meanings more often appear next to already important and frequently used concepts, rather than being distributed chaotically throughout the language.
Researchers from Fudan University, Harvard and Stony Brook University have studied how dictionaries of different languages are organised and how they have changed over time. To do this, they used natural language processing methods - that is, algorithms that know how to represent words as points in a semantic space.
Details
To put it simply, words that are close in meaning appear "next to each other" in this model. For example, words related to food, technology, power or family form their own semantic areas. The scientists checked how frequently used and new words are distributed in these areas.
It turned out that different languages have common patterns. Frequently used words gravitate to other frequently used words, forming a kind of "popular zones" of the dictionary. New words also do not appear evenly, but in groups - next to other recently emerged concepts.
This is similar to the way new topics appear in society: if a new technology, profession or cultural phenomenon develops, a whole set of new words quickly emerges around it.
Scientists have also found that languages have a similar hierarchical structure: words are grouped not only by individual topics, but also by larger semantic domains. These patterns emerged in all 22 languages that were included in the analysis.
Why it matters
The study shows that language does not just develop as a collection of random new words. Its growth is linked to what topics become important to society.
When culture, technology, politics or everyday life changes, language responds with bursts of new words and meanings. At the same time, similar processes occur in different languages, even if their speakers live in different countries and cultures.
This helps us to better understand not only the evolution of language, but also cultural history: the dictionary keeps traces of what was important to people in different eras.
Background
Previously, scientists already knew that statistical patterns operate in languages - for example, the most frequent words are used far more often than most others. The new work extends this approach: it shows not only the frequency of words, but also how words are distributed by meaning and time of occurrence.
The authors have also proposed a mathematical model that reproduces similar patterns. This could help study not only languages, but also other forms of cultural evolution - for example, the development of scientific ideas, technology or societal themes.
Source
The study by Xingzhi Guo and co-authors is published in Proceedings of the Royal Society B in 2026 under the title Statistical structure and the evolution of languages. They used data on English and 21 other languages, as well as NLP and statistical modelling techniques.