Automatic Analysis of school textbooks' syntactic complexity

The purpose of this study was to compare the syntactic complexity of texts used to communicate knowledge in the school textbooks of three school subjects. To do so, we collected a corpus of 2121 texts, used in the school textbooks that the State of Chile provides to students attending public schools. Texts were automatically analyzed by an algorithm that identifies syntactic dependency relations in a sentence and then calculates the mean Syntactic Dependency Length (SDL) of that sentence. Results showed that the SDL of the analyzed texts-corresponding to different levels and school subjects-was homogeneously low. Besides, it was possible to observe that there was not a pattern of incremental complexity associated with school levels. Results also showed that while it was not possible to identify disciplinary patterns that allowed the identification of school subjects exhibiting more CS, there was a tendency that places History, Geography and Social Science as the most syntactically complex.