N‐final relative clauses: The Amharic case

Abstract

Kayne develops a unique analysis for the relative clause construction based on the Linear Correspondence Axiom (LCA). According to Kayne (1994) outside the relative CP relative clauses do not have a nominal head. For him, the determiner directly selects the relative CP. Borsely (1997) and Platzack (1997/2000) argue that this kind of analysis has many drawbacks. However, their arguments come from N‐initial relative clause constructions. This paper examines how Kayne's approach handles N‐final relative clause constructions.We will immediately note that Kayne (1994) proposes the empirically erroneous generalisation that N‐final relative clauses lack an overt complementizer. Contrary to Kayne, languages such as Amharic clearly have a complementizer.This paper argues that from an empirical and theoretical point of view Kayne's proposal to N‐final relative clause construction has a lot of problems. In line with Platzack (1997/2000), this paper argues that N‐final relative clauses can be treated uniformly with N‐initial relative clauses as a complement of a “head” noun and moved to the functional definiteness Agr category to check the definiteness feature in the sense of Chomsky (1993).