How does language evolve as a multi-level system? A quantitative exploration of written Chinese

Hierarchy has been described as the backbone of a language system. However, how language evolves as a multi-level system has not been explored quantitatively based on authentic language materials. The Menzerath–Altmann law (MAL) is a statistical linguistic universal that can capture the complex relationships between language units at neighboring levels. Using the MAL, the present study explored the evolution of two regularly examined partial hierarchies in written Chinese, i.e., “clause-word-character” and “sentence-clause-word” across five periods of two millennia. The results indicate that the hierarchy in the Pre-Qin Period (Period 1) is quite different from the others since its linguistic units of character and word overlap to some extent. The two partial hierarchies show opposite evolutionary trends in the following four periods. The hierarchy fades at the “clause-word-character” levels. Nevertheless, it increases significantly at the “sentence-clause-word” levels. The evolutions are accompanied by a constant increase in word length and accelerated growth in clause length and sentence length/complexity. The findings are finally explained from the perspective of the Complex Adaptive System (CAS) theory.