We developed a multilingual model for readability. This model generates a score for Wikipedia articles capturing (some aspect) of how easy it is to read. For more details see: https://meta.wikimedia.org/wiki/Research:Multilingual_Readability_Research#An_improved_multilingual_model_for_readability
At the moment, the model lives on one of the stat-machines. The goal is to make the model's output available via Lifting.