Distance between Chinese registers based on the Menzerath-Altmann law and regression analysis

Abstract

This paper proposes an innovative method/index to represent the formality of a register based on the Menzerath–Altmann law and regression analysis. This index also can be used to quantify the distance between two registers. Analysis demonstrates that average word length decreases with the increase of clause length in each register and that their relationship can be fitted by the formula y = axb. It can be shown that the link between average word length and clause length abides by the Menzerath–Altmann law. Texts were represented by the fitted parameters, a and b, and their positions were plotted in 2-dimensions. Linear regression can be used to fit the functional correlation between these two parameters in each register. We show that the a-intercept of this regression line can be used as an index to represent the formality degree of the register and to compute the distance between two registers.