hirogari-tyuu: Category Extension in Japanese Lexical Classes and Parts of Speech

There are several ways to categorize the Japanese lexicon. Like many of the world’s languages, it is possible to divide Japanese into so-called lexical categories—within Japanese, nouns, verbs, adverbs, and adjectives are among those that are generally recognized. Additionally, there are some theories that claim the necessity of a “verbal noun” part of speech; however, these claims are not entirely convincing, and what exactly can count as a verbal noun requires specific elaboration. However, there are other methods for splitting up the lexicon—in particular, gosyu, or etymologically-based classes. In traditional Japanese grammar, the gosyu categories are Native (sometimes called Yamato), Sino-Japanese, and Foreign (that is, recent borrowings mostly from languages in western nations). Althought from a diachronic point of view, such categories may be necessary, there is room for argument as to their necessity in a synchronic grammar of Japanese. In this paper, I will examine a phenomenon that may shed some light on the workings of lexical categories (parts of speech) and gosyu. Specifically, I will examine the linguistic behavior of a form that expresses a temporal relationship, -tyuu, in relation to lexical classes and gosyu. Constructions that contain the temporal-relationship marker -tyuu or those like it (taking -tyuu as the prototypical example, I will refer to such constructions as “tyuuconstructions”) have been touched upon by Shibatani (1990) and Kageyama (1993), and in particular Miyamoto (1999) and Ohta (1995) have paid close attention to them. This paper will attempt to determine what sort of word or phrase is selected for by -tyuu, and what sort of characteristics those words or phrases must have. As will become clear later on, this question cannot be answered simply by examining lexical classes or gosyu, or even syntactic behavior surrounding some interaction of the two. I will show that semantic factors must play a crucial role. I then hope to further demonstrate that the primary factors are not merely “semantic,” but must be couched in the concepts of prototype effect and radial categories.