is added to the comparison to show how similar semantic motivations can lead to distinct (but related) grammatical encoding. We argue that the two languages illustrate two different ways languages can align the encoding of property concepts: aligning them... on the basis of one prototypical semantic property of verbs, or splitting such concepts into a first subclass on the basis of a prototypical semantic property of verbs and into a second subclass on the basis of a prototypical semantic property of nouns