This thesis investigates the grammatical properties and functions of Japanese mimetics when they are used as prenominal modifiers. I focus on the cases where mimetics modify nouns with physical referents. I argue that mimetic-na (M-na) should be considered neither ungrammatical nor less acceptable than other modifiers, contrary to suggestions in the previous literature. Looking at different grammatical markers combined with a mimetic, I demonstrate that M-na gives rise to a situation-descriptive reading, that mimetic-sita (M-sita) denotes a characterizing property and that mimetic-no (M-no) denotes a defining property, in Roy’s (2013) terms. The thesis includes examples in French, Russian and Spanish to illustrate these three different interpretations.
As for the syntactic structures of mimetic modifiers, I demonstrate that M-na is a tensed clausal modifier, while M-sita is a tenseless attributive modifier, following Hamano (1986, 1988, 1998). More specifically, I claim that M-sita is an AP. I provide evidence showing that M-na is tensed (allowing a temporally anchored interpretation), whereas M-sita disallows tensed interpretations. There is currently no consensus about the grammatical status of M-no. Based on the distributions of mimetic and non-mimetic words presented in this thesis, I suggest that M-no can be marked by either the genitive or the copula.
Each of the modifiers enters into a stacking structure when they occur together. I show that semantics associate with structural positions, and argue that mimetic modifiers appear in the order of M-na, M-sita, M-no in a hierarchical structure. This thesis sheds light on the various grammatical properties of mimetics in relation to their prosody. In broad agreement with previous research, I claim that accentless mimetics, as in M-na and M-no, denote an abstract quality, while I argue that M-sita (which involves an accented mimetic) denotes a physical concrete property. I consider the bare accented mimetics to be somewhat verb-like.