While proper nouns in natural languages are commonly assumed to be primitive
entities representing specific constants or individuals, recent literature (e.g., Matushansky 2008, Ghomeshi & Massam 2009, Izumi 2012, and Izumi 2016, among others) argues extensively that they are underlyingly predicative, and that their referential status is achieved compositionally. While this paper provides further evidence from Japanese that supports the predicate analysis of proper nouns, it also shows that Japanese exhibits a stark asymmetry between given names and family names, and thus it argues that proper nouns do not form a uniform class in syntax.