More than one diachronic corpus has shown that the Nominative-Genitive Conversion (NGC) in Japanese, which might appear to be a free alternation in adnominal clauses,has declined its frequency in the last 100 years or so (Nambu, 2014; Ogawa, 2016). Harada (1971, 1976) also identified two dialects roughly corresponding to two different age groups at the time; NCG in some specific syntactic environments was not acceptable for the Tokyo dialect speakers in their twenties, although it was acceptable for those in their forties. Primarily based on Kim's (2008) observations, Miyagawa (2011) proposes that synchronically, the more stative predicates are more likely to accept NGC. With these three previous analyses as backgrounds, this study administered an experiment of introspective judgments to 300 participants from three different age groups ranging from 74 to 20 years old. We discovered the fact that high stativity of an adnominal clause renders its Genitive subject more acceptable, and that younger participants are less likely to accept its Genitive subject. These two results enable us to conclude that the diachronic corpus investigation made by Ogawa (2016) is correct in that the restriction on NGC has been getting stronger and stronger and language change in this area is still in progress for the last 40 years. The generalization that more stative (or less eventive) predicates are more likely to accept NGC is to be attributed to the proposed hypothesis that more stative predicates need a smaller syntactic structure and that unlike the syntactic size of an adnominal clause with a Nominative subject, which is uniformly CP, the syntactic size of an adnominal clause with a Genitive subject has been shrinking from CP to VP/AP in the last 100 years or so.