出版社：Japan Science and Technology Information Aggregator, Electronic
摘要：In this paper, we are concerned with establishing the principles governing the choice between a wh-relative and a prep(osition)+which in time and place relative clauses (=RCs), such as (i) the day [{when/on which} ... [e]... ] and (ii) the street [[where/on which] ... [e]... ]. We argue, among other things, that the cases where a wh-relative is used, which we call “hamidashi” (or “pressed-out”) RCs, and those where a prep+which is used are two different types of RC. In “pressed-out” RCs the RC defines or characterizes the attribute of its antecedent and restricts the range of the attribute of the antecedent. This function differentiates “pressed-out” RCs from the “prep+which”-type (=restrictive RC). We claim that these semantic differences between the two types of RC can be attributed to the nature of the “gap” (indicated by “[e]” in (i) and (ii)) inside of the RC. We also suggest that there is a parallelism between pre-modification and post-modification, more specifically, that “pressed-out” RCs (postmodifiers) are comparable with classifying adjectives (pre-modifiers) in their relevant respects. Finally, we provide syntactic evidence that suggests that “pressedout” RCs are not complements: rather, both “pressed-out” RCs and ordinary restrictive RCs (including the “prep+which”-type) could be on the same level (both outside complement elements) in the X-bar structure. To reconcile such a situation with the claim that the “pressed-out” RC and the “prep+which”-type are two different types of RC, we propose a hypothesis to the effect that types of RC can be determined according to the nature of the gap regardless of their phrase-structural position.
