Network properties of written human language.

Area of Science:

• Computational Linguistics
• Network Science
• Literary Analysis

Background:

• Written human language can be analyzed using complex network theory.
• Understanding network topology, particularly local properties like nearest neighbors and clustering coefficients, is crucial.

Purpose of the Study:

• To investigate the network topology of written language, specifically George Orwell's "1984."
• To identify functional classes of vertices and essential network architecture components.
• To develop and validate a model that replicates empirical findings in language networks.

Main Methods:

• Complex network theory applied to the text of "1984."
• Analysis of local network properties: nearest neighbor degree and clustering coefficient.
• Extension of the Dorogovtsev and Mendes language model with accelerated growth mechanisms.

Main Results:

• Observed composite power law behavior in average nearest neighbor's degree and clustering coefficient.
• Identified distinct functional classes of vertices based on network properties.
• Demonstrated that second-order vertex correlations are integral to the network architecture.
• Developed an accelerated growing network model incorporating linear preferential attachment, local preferential attachment, and random growth.

Conclusions:

• The network architecture of written language exhibits composite power law behavior and significant second-order vertex correlations.
• The proposed accelerated growing network model effectively reproduces the empirical findings, generating syntactic-like structures.
• This research provides insights into the structural organization of language through the lens of complex networks.