Abstract
A number of previous studies have made an analysis of the adnominal usage of the Old Japanese auxiliary mu based on inductive methods; however, the essential problems have been left unsolved. This paper investigates and describes the adnominal usage of mu based on a deductive method. Specifically, this paper compares and describes the following two different types of noun phrases (NP): Type A: Inflected word+φ+hito (e.g., wakaki hito) Type B: Inflected word+mu+hito (e.g., wakakara-mu hito) An investigation of the syntactic features of these two types of NP in Japanese classics shows that Type B can be described as a noun phrase for non-reality; only Type B NP had restrictions in terms of types of predicates (i.e., they do not co-occur with existential predicates or predicates marked with tense/aspect forms), and co-occurence with expressions of time, place and quantity (with some exceptions which can be explained). The paper concludes that auxiliary mu in Old Japanese functioned as a marker indicating non-reality of the noun phrase.
