1. Introduction

Global language distribution is uneven, with varying degrees of language family development and significant internal differences. Language differences are the core of language comparative research, and traditional methods include subjective and objective measurements. Quantifying the differences within the world language family is challenging, but crucial for understanding language patterns and complexity. Swadesh proposed a method for calculating the percentage of cognates, while modern research such as the ASJP database analyzes language relationships through lexical distance analysis. This article uses the ASJP database to calculate the similarity of 20 major language families and study their internal differences.

2. Language materials and method explanations

Discussed the sources and research methods of language materials. Ethnography and Glottolog record the types and language families in the world, with Ethnography recording 7164 languages divided into 156 language families, while Glottolog records 8604 languages divided into 430 language families. The study focuses on 20 major language families, which cover 83.02% of the world's language population. The study utilized linguistic materials from the ASJP database and employed a language similarity calculation program based on the ASJP model, particularly the Normalized Lewinstein Distance Quotient (LDND), to compute the similarity between languages. In the data analysis section, violin plots were created using GraphPad Prism software to display the distribution status and probability density of similarity values between each two languages, in order to reflect the magnitude of internal differences within the languages.

3. Differences within the 20 major language families in the world

Explored the internal differences of 20 major language families in the world, pointing out that the inconsistency and complexity of language evolution lead to significant differences within each language family. By using the language similarity index proposed by Weishiman and Ran Qibin, the language similarity of these language families was analyzed, and it was found that the average was between 2.42 and 34.76. The differences within the Pan New Guinea language family are the greatest, while the differences within the Maya language family are the smallest. Through quartile analysis, the particularity of language differences between the Otomang language family and the Utah Aztec language family was revealed. In addition, the distribution characteristics of similarity data among different language families were shown through violin plots, including symmetrical "olive shaped", asymmetrical "olive shaped", and "dispersed" distributions, reflecting the linguistic diversity and internal structure of different language families.

4. Analysis of internal differences in some major language families

Since William Jones proposed the phylogenetic relationship between Indo European languages in 1786, the study of Indo European language families has received widespread attention. The 20 major language families in the world contain 83.02% of languages, but there are differences in classification methods and results. The Sino Tibetan language family is one of the largest and most diverse language families, including various morphological system types. There are significant differences within the Sino Tibetan language family, with the largest differences between the Tibeto Burman language family and the smallest differences between the Chinese language family. The Altaic language family has high internal similarity, with the smallest differences in the Turkic language family and the largest differences in the Mongolian language family. In the Nile Saharan language family, East Sudanese has significant differences and low similarity with other languages. In the Niger Congo language family, the differences between Mandeb and other languages are close to the critical value of the same language family, but the overall internal differences are significant.

5. Conclusion

The language similarity of 20 major language families was analyzed using the ASJP model, and it was found that the Pan New Guinea language family had the largest internal differences, while the Maya language family had the smallest. The Tibeto Burman language family has the greatest differences in the Sino Tibetan language family, while the Chinese language family has the smallest differences. The data distribution shows that symmetric "olive shaped" is common in language families with large differences, asymmetric "olive shaped" appears in language families with moderate differences, and "dispersed" is more common in language families with small differences. The perspective of language similarity provides a new explanation for the controversial classification of language families. The research results contribute to understanding the language relationships and evolutionary processes within different language families, demonstrating the effectiveness of the ASJP model in language synchronic and diachronic studies.