AlphaDepLength: A New Measure of Syntactic Complexity Based on Communicative Efficiency and Its Prediction of Reading Time
Lei Lei, Yaqian Shi · Journal of Quantitative Linguistics · 2025
Previous studies have primarily focused on communicative efficiency at the lexical level, with less attention paid to the syntactic level due to the lack of a valid measure of syntactic complexity that reflects human processing of syntactic structures. To address the issue, this study proposes a new measure of syntactic complexity based on the theory of communicative efficiency. It hypothesizes that when humans hierarchically process a sentence, they tend to choose the shortest mean hierarchical length to minimize their efforts. Therefore, the mean hierarchical length calculated from all hierarchical lengths in a sentence may determine its syntactic complexity. Based on such a hypothesis, we propose a new measure of syntactic complexity, i.e. the mean hierarchical dependency length (AlphaDepLength). To examine its validity, we conducted an experiment to compare the performance of AlphaDepLength with that of other syntactic measures such as mean dependency distance (MDD), mean hierarchical distance (MHD), and mean hierarchical dependency distance (MHDD) in reading time prediction. Results show that AlphaDepLength has a stronger correlation with reading time and explains more variance in reading time prediction. The results indicate that AlphaDepLength may well reflect the syntactic processing in our mind and thus it is a better proxy of syntactic complexity of a sentence.