EUVIMEDEUVIMED entdecken
Crossref

An empirical study on the complex network characteristics of the Japanese language

Huimin Xu, Junwei Liu

Digital Scholarship in the Humanities · 20.08.2026

DOI: 10.1093/llc/fqag116 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Digital Scholarship in the Humanities, online erschienen am 20.08.2026. Verfasst von Huimin Xu und 1 weiteren Autor:innen. Die Arbeit nennt 24 Literaturangaben. Gefördert von: Social Science Foundation of Shandong Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study selects eight registers of modern Japanese: interview, speech, official document, sci-tech text, political commentary, essay, poem, and novel, as corpora. Undirected and unweighted Japanese word co-occurrence networks are constructed using MeCab and Gephi. Based on complex network theory, this paper investigates the small-world and scale-free properties of Japanese linguistic networks. It applies quantitative tools: Erdős-Rényi random network benchmark, small-world index S, two-way analysis of variance, nonlinear fitting, piecewise regression, and Bootstrap resampling. Power-law tests follow the Clauset et al. method. Results indicate network size and register type exert significant main and interaction effects on small-world strength. The theoretically critical node number for small-world emergence is ninety-one, the empirical threshold eighty-nine, both falling within the 95 per cent confidence interval (81–146), confirming consistent theoretical and empirical findings. However, after removing high-frequency function words such as particles and auxiliary verbs, the network S-values are generally lower than 3, the threshold effect disappears, and the small-world property becomes unstable. The small-world property is neither an inherent attribute of the Japanese lexical system nor a direct outcome of the internal structure of word co-occurrence; instead, it heavily relies on the grammatical scaffolding constructed by high-frequency function words serving as hub nodes. Of the 119 sampled networks, 88 per cent pass power-law tests and show standard hierarchical scale-free features: core words form dense hubs, while peripheral words have sparse connections. There exists a moderately weak yet statistically significant correlation between the number of nodes and the power-law exponent. This research provides theoretical and methodological support for stylometry, Japanese structural linguistic analysis, and natural language processing.

Förderung

  • Social Science Foundation of Shandong Province

Zitieren

APA

Xu, H., & Liu, J. (2026). An empirical study on the complex network characteristics of the Japanese language. Digital Scholarship in the Humanities. https://doi.org/10.1093/llc/fqag116

Vancouver

Xu H, Liu J. An empirical study on the complex network characteristics of the Japanese language. Digital Scholarship in the Humanities. 2026. doi:10.1093/llc/fqag116

BibTeX
@article{Xu2026,
  author = {Xu, Huimin and Liu, Junwei},
  title = {An empirical study on the complex network characteristics of the Japanese language},
  journal = {Digital Scholarship in the Humanities},
  year = {2026},
  doi = {10.1093/llc/fqag116}
}
Kostenlos starten →