Abstract:As a complex adaptive system,language abides by certain laws in the course of evolution and development,which can be investigated through quantitative approaches.We use a large English dependency treebank to analyze the hierarchical structure of English sentences.Some particular measures such as vertice number (VN),hierarchical number (HN),hierarchical distance (HD),mean hierarchical distance (MHD) and English mean hierarchical distance (MHD ₂) are introduced to evaluate the hierarchical structure of English sentences.We observe that there are significantly positive correlations between VN,HN,and MHD.A positive skew is also found in the distribution of HN,which indicates a preference for flatter structures in English sentences,and the skewness coefficient is increasing for more vertices.Furthermore,we propose a threshold below 4 for English MHD ₂.