Clausal Complement or Adverbial Clause?: Toward an Integral Account of the Japanese Internally-Headed Relative Clause

This paper proposes a constraint-based, comprehensive analysis of the Internally-Headed Relative Clause (henceforth, IHRC) in Japanese which is able to accommodate the synchronic properties as well as the diachronic change that IHRC went through. Various types of syntactic analyses have been proposed to date to explain the peculiar traits of IHRC as it is today. Philological studies in Japanese have documented the gradual historical change in the system of Japanese that gave rise to IHRC. With a few exceptions such as Kuroda (1975-77), syntactic studies have mainly focused on the synchronic aspect of IHRC, and have not attempted to unify the two perspectives to draw a more complete picture of IHRC within the system of the Japanese language. This paper claims that IHRC should be defined as a special type of non-head daughter of a head-complement structure. It explicitly denies that IHRC in isolation designates a certain entity (or a target). The propsed analysis not only accounts for the peculiar traits of IHRC in today’s Japanese, but provides a rational explanation for the process of grammatical change that IHRC has gone through. IHRC has been observed in various genetically unrelated languages including Navaho, Lakhota, Dogon, Quechua, as well as in Japanese and Korean, although the Japanese one (and plausibly Korean ones as well) is deemed rather exceptional by Culy (1990). The defining characteristic of IHRC, according to Culy (1990) is (1) that it has the basic structure of a sentence, (2) that it refers to a noun which occurs internal to the clause, and (3) that the semantic head (=target) typically bears no syntactic or morphological marker as such. Thus IHRC by definition presents challenge in reconciling the apparent type conflict of NP and S, and in finding a proper mechanism of identifying the target. The first point of challenge motivated Culy (1990) to propose an exocentric structure where an N-bar dominates an S, while the second motivated him to rely on a coindexation mechanism common to both IHRC and EHRC (i.e. the usual type of relative clause.) Japanese IHRC apparently share all of the above definining properties. This is not surprising, given that Japanese also meets the generalizations given by Cole (1987) and Culy (1990) of those