It's very interesting. Thank you for making it open-source. Would you like to explain the inner workings of ELD? For example, what is the model of language that it produces, how does it compare a new word to the model and how is the word scored for different languages.